Professional-Machine-Learning-Engineer Google Professional Machine Learning Engineer Questions and Answers

The F1 score is a metric that combines both precision and recall, and is suitable for evaluating imbalanced classification problems. Precision measures the fraction of true positives among the predicted positives, and recall measures the fraction of true positives among the actual positives. The F1 score is the harmonic mean of precision and recall, and it ranges from 0 to 1, with higher values indicating better performance. The F1 score is a good metric for imbalanced data because it balances both the false positives and the false negatives, and does not favor the majority class over the minority class.

The other options are not good metrics for imbalanced data. RMSE (root mean squared error) is a metric for regression problems, not classification problems. It measures the average squared difference between the predicted and the actual values, and is not suitable for binary outcomes. F score with higher precision weighting than recall, or F0.5 score, is a metric that gives more importance to precision than recall. This means that it penalizes false positives more than false negatives, which is not desirable for imbalanced data where the minority class is more important. F score with higher recall weighting than precision, or F2 score, is a metric that gives more importance to recall than precision. This means that it penalizes false negatives more than false positives, which might be suitable for some imbalanced data problems, but not for the logo detection problem. In this problem, both false positives and false negatives are equally important, as we want to accurately identify the presence or absence of a logo in an image. Therefore, the F1 score is a better metric than the F2 score. References:

• Tour of Evaluation Metrics for Imbalanced Classification
• Metrics for imbalanced data (simply explained)

According to the official exam guide1, one of the skills assessed in the exam is to “design, build, and productionalize ML models to solve business challenges using Google Cloud technologies”. Cloud Data Loss Prevention (DLP) API2 is a service that provides programmatic access to a powerful detection engine for personally identifiable information and other privacy-sensitive data in unstructured data streams, such as text blocks and images. Cloud DLP API helps you discover, classify, and protect your sensitive data by using techniques such as de-identification, masking, tokenization, and bucketing. You can use Cloud DLP API to de-identify the PII data before performing data exploration and preprocessing, and retain the data utility for ML purposes. Therefore, option A is the best way to perform data exploration and preprocessing while ensuring the security and privacy of sensitive fields. The other options are not relevant or optimal for this scenario. References:

• Professional ML Engineer Exam Guide
• Cloud Data Loss Prevention (DLP) API
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

According to the official exam guide1, one of the skills assessed in the exam is to “explain the predictions of a trained model”. Vertex AI provides feature attributions using Shapley Values, a cooperative game theory algorithm that assigns credit to each feature in a model for a particular outcome2. Feature attributions can help you understand how the model calculates the predictions and debug or optimize the model accordingly. You can use Forecasting with AutoML or Tabular Workflow for Forecasting to generate and query local feature attributions2. The other options are not relevant or optimal for this scenario. References:

• Professional ML Engineer Exam Guide
• Feature attributions for forecasting
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

According to the web search results, Reduction Server is a faster GPU all-reduce algorithm developed at Google that uses a dedicated set of reducers to aggregate gradients from workers12. Reducers are lightweight CPU VM instances that are significantly cheaper than GPU VMs2. Therefore, the third worker pool should not have any accelerators, and should use a machine type that has high network bandwidth to optimize the communication between workers and reducers2. TPUs are not supported by Reduction Server, so the first two worker pools should have GPUs and use a container image that contains the training code12. The reduction-server container image is provided by Google and should be used for the third worker pool2.

 TensorFlow Data Validation (TFDV) is a library that helps you understand, validate, and monitor your data for machine learning. It can automatically detect and report schema anomalies, such as missing features, new features, or different data types, in your data. It can also generate descriptive statistics and data visualizations to help you explore and debug your data. TFDV can be integrated with your model training pipeline to ensure data quality and consistency throughout the machine learning lifecycle. References:

• TensorFlow Data Validation
• Data Validation | TensorFlow
• Data Validation | Machine Learning Crash Course | Google Developers

This is not a good result because the model is performing worse than predicting that people will always renew their subscription. This option has the following reasons:

• It indicates that the model is not learning from the data, but rather memorizing the majority class. Since 90% of the individuals renew their subscription every year, the model can achieve a 90% accuracy by simply predicting that everyone will renew their subscription, without considering the features or the patterns in the data. However, the model’s accuracy for predicting those who renew their subscription is only 82%, which is lower than the baseline accuracy of 90%. This suggests that the model is overfitting to the minority class (those who cancel their subscription), and underfitting to the majority class (those who renew their subscription).
• It implies that the model is not useful for the business problem, as it cannot identify the customers who are at risk of churning. The goal of predicting whether customers will cancel their annual subscription is to prevent customer churn and increase customer retention. However, the model’s accuracy for predicting those who cancel their subscription is 99%, which is too high and unrealistic, as it means that the model can almost perfectly identify the customers who will churn, without any false positives or false negatives. This may indicate that the model is cheating or exploiting some leakage in the data, such as a feature that reveals the outcome of the prediction. Moreover, the model’s accuracy for predicting those who renew their subscription is 82%, which is too low and unreliable, as it means that the model can miss many customers who will churn, and falsely label them as renewing customers. This can lead to losing customers and revenue, and failing to take proactive actions to retain them.

References:

• How to Evaluate Machine Learning Models: Classification Metrics | Machine Learning Mastery
• Imbalanced Classification: Predicting Subscription Churn | Machine Learning Mastery

According to the web search results, Cloud Build1 is a service that executes your builds on Google Cloud Platform infrastructure. Cloud Build can import source code from Cloud Source Repositories2, Cloud Storage, GitHub, Bitbucket, or any publicly hosted Git repository. Cloud Build allows you to create and manage build triggers, which are automated workflows that run whenever a code change is pushed to your source repository. You can use Cloud Build triggers to automatically retrain your ML models whenever there is any modification of the code. Therefore, option B is the best way to set up the CI pipeline for the given use case, as it allows you to configure a Cloud Build trigger with the event set as “Push to a branch”, which means the trigger will run whenever a new commit is pushed to a specific branch of your source repository. The other options are not relevant or optimal for this scenario. References:

• Cloud Build
• Cloud Source Repositories
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

 Entity extraction is a natural language processing (NLP) task that involves identifying and extracting specific types of information from text, such as names, dates, locations, etc. Entity extraction can help you analyze a corpus of recipes and extract each ingredient and cookware mentioned in them. Vertex AI is a unified platform for building and managing machine learning solutions on Google Cloud. It provides a service for AutoML entity extraction, which allows you to create and train custom entity extraction models without writing any code. You can use Vertex AI to create a text dataset for entity extraction, and label your data with two entities: “ingredient” and “cookware”. You need to label at least 200 examples of each entity type to train an AutoML entity extraction model. You can also use a holdout dataset to evaluate the performance of your model, such as precision, recall, and F1-score. This solution can help you build a machine learning model to scan a corpus of recipes and extract each ingredient and cookware mentioned in them, and use the results to help users with meal planning. References:

• AutoML Entity Extraction | Vertex AI
• Preparing data for AutoML Entity Extraction | Vertex AI

NLTK is a Python library that provides a set of tools for natural language processing, such as tokenization, stemming, tagging, parsing, and sentiment analysis. Tokenization is a process of breaking a text into smaller units, such as words or sentences. You can use NLTK to quickly experiment with tokenizing text in a managed Vertex AI Workbench notebook. A Vertex AI Workbench notebook is a web-based interactive environment that allows you to write and execute Python code on Google Cloud. You can install the NLTK library from a Jupyter cell by using the !pip install nltk --user command. This command uses the pip package manager to install the NLTK library for the current user. By installing the NLTK library from a Jupyter cell, you can avoid the hassle of opening a terminal or creating a custom image for your notebook. References:

• NLTK documentation
• Vertex AI Workbench documentation
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

BigQuery ML is a service that allows you to create and train ML models using SQL queries. You can use BigQuery ML to train an AutoML regression model, which is a type of model that automatically selects the best features and architecture for your data. You can also specify Vertex AI as the model registry, which is a service that allows you to store and manage your ML models. By using Vertex AI as the model registry, you can easily deploy your model to a Vertex AI endpoint, which is a service that allows you to serve your ML models online and scale them automatically. By using BigQuery ML, Vertex AI model registry, and Vertex AI endpoint, you can deploy your model for online prediction as quickly as possible, without having to export, import, or retrain your model. References:

• BigQuery ML documentation
• Vertex AI documentation
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

 XGBoost is a popular open-source library that provides a scalable and efficient implementation of gradient boosted trees. You can use XGBoost to train a classification or regression model on tabular data. You can also use Vertex AI to productionize the model and expose it for internal use as an HTTP microservice. Vertex AI is a service that allows you to create and train ML models using Google Cloud technologies. You can use a prebuilt XGBoost Vertex container to create a model and deploy it to Vertex AI Endpoints. A prebuilt Vertex container is a container image that contains the dependencies and libraries needed to run a specific ML framework, such as XGBoost. You can use a prebuilt Vertex container to simplify the model creation and deployment process, without having to build your own custom container. Vertex AI Endpoints is a service that allows you to serve your ML models online and scale them automatically. You can use Vertex AI Endpoints to deploy the model from the prebuilt Vertex container and expose it as an HTTP microservice. You can also configure the endpoint to handle a small number of incoming requests, and optimize the latency and cost of serving the model. By using a prebuilt XGBoost Vertex container and Vertex AI Endpoints, you can productionize the model with the least amount of effort and latency. References:

• XGBoost documentation
• Vertex AI documentation
• Prebuilt Vertex container documentation
• Vertex AI Endpoints documentation
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

 The best option for monitoring the model to determine when retraining is necessary is to schedule a weekly query in BigQuery to compute the success metric. This option has the following advantages:

• It allows the model performance to be evaluated regularly, based on the actual outcome of the recommendations. By computing the success metric, which is the percentage of articles that are opened within two days and read for at least one minute, you can measure how well the model is achieving its objective and compare it with the acceptable baseline.
• It leverages the scalability and efficiency of BigQuery, which is a serverless, fully managed, and highly scalable data warehouse that can run complex queries over petabytes of data in seconds. By using BigQuery, you can access and analyze all the information needed to compute the success metric, such as the newsletter publication date, the article opening date, and the user reading time, without worrying about the infrastructure or the cost.
• It simplifies the model monitoring and retraining workflow, as the weekly query can be scheduled and executed automatically using BigQuery’s built-in scheduling feature. You can also set up alerts or notifications to inform you when the success metric falls below the acceptable baseline, and trigger the model retraining process accordingly.

The other options are less optimal for the following reasons:

• Option A: Using Vertex AI Model Monitoring to detect skew of the input features with a sample rate of 100% and a monitoring frequency of two days introduces additional complexity and overhead. This option requires setting up and managing a Vertex AI Model Monitoring service, which is a managed service that provides various tools and features for machine learning, such as training, tuning, serving, and monitoring. However, using Vertex AI Model Monitoring to detect skew of the input features may not reflect the actual performance of the model, as skew is the discrepancy between the distributions of the features in the training dataset and the serving data, which may not affect the outcome of the recommendations. Moreover, using a sample rate of 100% and a monitoring frequency of two days may incur unnecessary cost and latency, as it requires analyzing all the input features every two days, which may not be needed for the model monitoring.
• Option B: Scheduling a cron job in Cloud Tasks to retrain the model every week before the newsletter is created introduces additional cost and risk. This option requires creating and running a cron job in Cloud Tasks, which is a fully managed service that allows you to schedule and execute tasks that are invoked by HTTP requests. However, using Cloud Tasks to retrain the model every week may not be optimal, as it may retrain the model more often than necessary, wasting compute resources and cost. Moreover, using Cloud Tasks to retrain the model before the newsletter is created may introduce risk, as it may deploy a new model version that has not been tested or validated, potentially affecting the quality of the recommendations.
• Option D: Scheduling a daily Dataflow job in Cloud Composer to compute the success metric introduces additional complexity and cost. This option requires creating and running a Dataflow job in Cloud Composer, which is a fully managed service that runs Apache Airflow pipelines for workflow orchestration. Dataflow is a fully managed service that runs Apache Beam pipelines for data processing and transformation. However, using Dataflow and Cloud Composer to compute the success metric may not be necessary, as it may add more steps and overhead to the model monitoring process. Moreover, using Dataflow and Cloud Composer to compute the success metric daily may not be optimal, as it may compute the success metric more often than needed, consuming more compute resources and cost.

References:

• [BigQuery documentation]
• [Vertex AI Model Monitoring documentation]
• [Cloud Tasks documentation]
• [Cloud Composer documentation]
• [Dataflow documentation]

To configure a model retraining policy that minimizes cost, you should follow these steps:

• Download the weather data each week, and download the flu data each month. This way, you can keep your data up to date with the latest information available, without downloading unnecessary or redundant data.
• Deploy the model to a Vertex AI endpoint with feature drift monitoring. This feature allows you to detect when the distribution of the input data changes significantly from the training data, which could affect the model performance1.
• Retrain the model if a monitoring alert is detected. This way, you can update your model only when needed, instead of retraining it on a fixed schedule, which could incur more cost and time.

References:

• 1: Monitor models for feature drift | Vertex AI | Google Cloud

The best option for handling missing data in this case is to predict the missing values using linear regression. Linear regression is a supervised learning technique that can be used to estimate the relationship between a continuous target variable and one or more predictor variables. In this case, the target variable is the distance from the closest school, and the predictor variables are the other features in the dataset, such as house size, location, number of rooms, etc. By fitting a linear regression model on the data that has no missing values, we can then use the model to predict the missing values for the distance from the closest school feature. This way, we can preserve all the instances in the dataset and avoid introducing bias or reducing variance. The other options are not suitable for handling missing data in this case, because:

• Deleting the rows that have missing values would reduce the size of the dataset and potentially lose important information. Since every instance is important, we want to keep as much data as possible.
• Applying feature crossing with another column that does not have missing values would create a new feature that combines the values of two existing features. This might increase the complexity of the model and introduce noise or multicollinearity. It would not solve the problem of missing values, as the new feature would still have missing values whenever the distance from the closest school feature is missing.
• Replacing the missing values with zeros would distort the distribution of the feature and introduce bias. It would also imply that the houses with missing values are located at the same distance from the closest school, which is unlikely to be true. A zero value might also be outside the range of the feature, as the distance from the closest school is unlikely to be exactly zero for any house. References:
• Linear Regression
• Imputation of missing values
• Google Cloud launches machine learning engineer certification
• Google Professional Machine Learning Engineer Certification
• Professional ML Engineer Exam Guide
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

 Customer churn is a binary classification problem, where the target variable is whether a customer has churned or not. Therefore, a logistic regression model is more suitable than a linear regression model, which is used for regression problems. A logistic regression model can output the probability of a customer churning, which can be used to rank the customers by their churn risk and take appropriate actions1.

BigQuery ML is a service that allows you to create and execute machine learning models in BigQuery using standard SQL queries2. You can use BigQuery ML to create a logistic regression model for customer churn prediction by using the CREATE MODEL statement and specifying the LOGISTIC_REG model type3. You can use the historical customer data as the input table for the model, and specify the features and the label columns3.

Vertex AI Model Registry is a central repository where you can manage the lifecycle of your ML models4. You can import models from various sources, such as BigQuery ML, AutoML, or custom models, and assign them to different versions and aliases4. You can also deploy models to endpoints, which are resources that provide a service URL for online prediction.

By registering the BigQuery ML model in Vertex AI Model Registry, you can leverage the Vertex AI features to evaluate and monitor the model performance4. You can use Vertex AI Experiments to track and compare the metrics of different model versions, such as accuracy, precision, recall, and AUC. You can also use Vertex AI Explainable AI to generate feature attributions that show how much each input feature contributed to the model’s prediction.

The other options are not suitable for your scenario, because they either use the wrong model type, such as linear regression, or they do not use Vertex AI to evaluate the model performance, which would limit the insights and actions you can take based on the model results.

References:

• Logistic Regression for Machine Learning
• Introduction to BigQuery ML | Google Cloud
• Creating a logistic regression model | BigQuery ML | Google Cloud
• Introduction to Vertex AI Model Registry | Google Cloud
• [Deploy a model to an endpoint | Vertex AI | Google Cloud]
• [Vertex AI Experiments | Google Cloud]

 Oversampling is a technique for dealing with imbalanced datasets, where the majority class dominates the minority class. It balances the distribution of classes by increasing the number of samples in the minority class. Oversampling can improve the performance of a classifier by reducing the bias towards the majority class and increasing the sensitivity to the minority class.

In this case, the dataset includes transactions, of which 1% are identified as fraudulent. This means that the fraudulent transactions are the minority class and the non-fraudulent transactions are the majority class. A random forest model trained on this dataset might have a low recall for the fraudulent transactions, meaning that it might miss many of them and fail to detect fraud. This could have a high cost for the bank and its customers.

One way to overcome this problem is to oversample the fraudulent transactions 10 times, meaning that each fraudulent transaction is duplicated 10 times in the training dataset. This would increase the proportion of fraudulent transactions from 1% to about 10%, making the dataset more balanced. This would also make the random forest model more aware of the patterns and features that distinguish fraudulent transactions from non-fraudulent ones, and thus improve its accuracy and recall for the minority class.

For more information about oversampling and other techniques for imbalanced data, see the following references:

• Random Oversampling and Undersampling for Imbalanced Classification
• Exploring Oversampling Techniques for Imbalanced Datasets

A random forest is an ensemble learning method that consists of many decision trees. It can be used for both regression and classification tasks. A random forest classification model can predict the probability of churn for each customer by assigning them to different classes, such as high-risk, medium-risk, or low-risk. A random forest model can also generate feature importances, which measure how much each feature contributes to the prediction. Feature importances can help interpret the model and understand what factors influence customer churn. Vertex AI Workbench is an integrated development environment (IDE) that allows you to create and run Jupyter notebooks on Google Cloud. You can use Vertex AI Workbench to build a random forest classification model in Python, using libraries such as scikit-learn or TensorFlow. You can also configure the model to generate feature importances after the model is trained, and visualize them using plots or tables. This solution can help you build an interpretable model for customer churn prediction, and use the results to design marketing campaigns that target at-risk customers. References:

• Random Forests | scikit-learn
• Vertex AI Workbench | Google Cloud
• Interpreting Random Forests | Towards Data Science

According to the web search results, Vertex AI Pipelines is a serverless orchestrator for running ML pipelines, using either the KFP SDK or TFX1. Vertex AI Pipelines provides a set of prebuilt components that can be used to perform common ML tasks, such as training, evaluation, deployment, and more2. Vertex AI ModelUploadOp and ModelDeployOp are two such components that can be used to upload a new version of the XGBoost model to Vertex AI Model Registry and deploy it to Vertex AI Endpoints for online inference3. Therefore, option D is the best way to use the simplest approach for the given use case, as it only requires chaining two prebuilt components together. The other options are not relevant or optimal for this scenario. References:

• Vertex AI Pipelines
• Google Cloud Pipeline Components
• Vertex AI ModelUploadOp and ModelDeployOp
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

 AutoML Natural Language is a service that allows you to build and train custom natural language models without writing code. You can use AutoML Natural Language to perform sentiment analysis with custom categories, such as positive, negative, or neutral. You can also use pre-trained models or transfer learning to leverage existing knowledge and reduce the amount of data required to train a model from scratch. AutoML Natural Language provides a user-friendly interface and a powerful AutoML engine that optimizes your model for high predictive performance.

Cloud Natural Language API is a service that provides pre-trained models for common natural language tasks, such as sentiment analysis, entity analysis, and syntax analysis. However, it does not allow you to customize the categories or use your own data for training.

AI Hub pre-made Jupyter Notebooks are interactive documents that contain code, text, and visualizations for various machine learning scenarios. However, they require some coding skills and data preparation to use them effectively.

AI Platform Training built-in algorithms are pre-configured machine learning algorithms that you can use to train models on AI Platform. However, they do not support sentiment analysis as a natural language task.

References:

• AutoML Natural Language documentation
• Cloud Natural Language API documentation
• AI Hub documentation
• AI Platform Training documentation

The best option for choosing a model that prioritizes detection while ensuring that more than 50% of the maintenance jobs triggered by the model address an imminent machine failure is to choose the model with the highest recall where precision is greater than 0.5. This option has the following advantages:

• It maximizes the recall, which is the proportion of actual failures that are correctly predicted by the model. Recall is also known as sensitivity or true positive rate (TPR), and it is calculated as:

mathrmRecall=fracmathrmTPmathrmTP+mathrmFN

where TP is the number of true positives (actual failures that are predicted as failures) and FN is the number of false negatives (actual failures that are predicted as non-failures). By maximizing the recall, the model can reduce the number of false negatives, which are the most costly and undesirable outcomes for the predictive maintenance use case, as they represent missed failures that can lead to machine breakdown and downtime.

• It constrains the precision, which is the proportion of predicted failures that are actual failures. Precision is also known as positive predictive value (PPV), and it is calculated as:

mathrmPrecision=fracmathrmTPmathrmTP+mathrmFP

where FP is the number of false positives (actual non-failures that are predicted as failures). By constraining the precision to be greater than 0.5, the model can ensure that more than 50% of the maintenance jobs triggered by the model address an imminent machine failure, which can avoid unnecessary or wasteful maintenance costs.

The other options are less optimal for the following reasons:

• Option A: Choosing the model with the highest area under the receiver operating characteristic curve (AUC ROC) and precision greater than 0.5 may not prioritize detection, as the AUC ROC does not directly measure the recall. The AUC ROC is a summary metric that evaluates the overall performance of a binary classifier across all possible thresholds. The ROC curve plots the TPR (recall) against the false positive rate (FPR), which is the proportion of actual non-failures that are incorrectly predicted by the model. The AUC ROC is the area under the ROC curve, and it ranges from 0 to 1, where 1 represents a perfect classifier. However, choosing the model with the highest AUC ROC may not maximize the recall, as the AUC ROC is influenced by both the TPR and the FPR, and it does not account for the precision or the specificity (the proportion of actual non-failures that are correctly predicted by the model).
• Option B: Choosing the model with the lowest root mean squared error (RMSE) and recall greater than 0.5 may not prioritize detection, as the RMSE is not a suitable metric for binary classification. The RMSE is a regression metric that measures the average magnitude of the error between the predicted and the actual values. The RMSE is calculated as:

mathrmRMSE=sqrtfrac1nsumi=1n​(yi​−hatyi​)2

where yi​ is the actual value, hatyi​ is the predicted value, and n is the number of observations. However, choosing the model with the lowest RMSE may not optimize the detection of failures, as the RMSE is sensitive to outliers and does not account for the class imbalance or the cost of misclassification.

• Option D: Choosing the model with the highest precision where recall is greater than 0.5 may not prioritize detection, as the precision may not be the most important metric for the predictive maintenance use case. The precision measures the accuracy of the positive predictions, but it does not reflect the sensitivity or the coverage of the model. By choosing the model with the highest precision, the model may sacrifice the recall, which is the proportion of actual failures that are correctly predicted by the model. This may increase the number of false negatives, which are the most costly and undesirable outcomes for the predictive maintenance use case, as they represent missed failures that can lead to machine breakdown and downtime.

References:

• Evaluation Metrics (Classifiers) - Stanford University
• Evaluation of binary classifiers - Wikipedia
• Predictive Maintenance: The greatest benefits and smart use cases

A ResourceExhaustedError: out of memory (OOM) when allocating tensor is an error that occurs when the GPU runs out of memory while trying to allocate memory for a tensor. A tensor is a multi-dimensional array of numbers that represents the data or the parameters of a machine learning model. The size and shape of a tensor depend on various factors, such as the input data, the model architecture, the batch size, and the optimization algorithm1.

For the use case of training a computer vision model that predicts the type of government ID present in a given image using a GPU-powered virtual machine on Compute Engine, the best option to resolve the error is to reduce the batch size. The batch size is a parameter that determines how many input examples are processed at a time by the model. A larger batch size can improve the model’s accuracy and stability, but it also requires more memory and computation. A smaller batch size can reduce the memory and computation requirements, but it may also affect the model’s performance and convergence2.

By reducing the batch size, the GPU can allocate less memory for each tensor, and avoid running out of memory. Reducing the batch size can also speed up the training process, as the GPU can process more batches in parallel. However, reducing the batch size too much may also have some drawbacks, such as increasing the noise and variance of the gradient updates, and slowing down the convergence of the model. Therefore, the optimal batch size should be chosen based on the trade-off between memory, computation, and performance3.

The other options are not as effective as option B, because they are not directly related to the memory allocation of the GPU. Option A, changing the optimizer, may affect the speed and quality of the optimization process, but it may not reduce the memory usage of the model. Option C, changing the learning rate, may affect the convergence and stability of the model, but it may not reduce the memory usage of the model. Option D, reducing the image shape, may reduce the size of the input tensor, but it may also reduce the quality and resolution of the image, and affect the model’s accuracy. Therefore, option B, reducing the batch size, is the best answer for this question.

References:

• ResourceExhaustedError: OOM when allocating tensor with shape - Stack Overflow
• How does batch size affect model performance and training time? - Stack Overflow
• How to choose an optimal batch size for training a neural network? - Stack Overflow

 The best option for performing hyperparameter tuning on Vertex AI to determine the appropriate embedding dimension and the optimal learning rate is to use UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learning rate, and a large number of parallel trials. This option has the following advantages:

• It matches the appropriate scaling type for each hyperparameter, based on their range and distribution. The embedding dimension is an integer hyperparameter that varies linearly between 16 and 64, so using UNIT_LINEAR_SCALE makes sense. The learning rate is a double hyperparameter that varies exponentially between 10e-05 and 10e-02, so using UNIT_LOG_SCALE is more suitable.
• It maximizes the exploration of the hyperparameter space, by using a large number of parallel trials. Since training time is not a concern, using more trials can help find the best combination of hyperparameters that maximizes model accuracy. The default Bayesian optimization tuning algorithm can efficiently sample the hyperparameter space and converge to the optimal values.

The other options are less optimal for the following reasons:

• Option B: Using UNIT_LINEAR_SCALE for the embedding dimension, UNIT_LOG_SCALE for the learning rate, and a small number of parallel trials, reduces the exploration of the hyperparameter space, by using a small number of parallel trials. Since training time is not a concern, using fewer trials can miss some potentially good combinations of hyperparameters that maximize model accuracy. The default Bayesian optimization tuning algorithm can benefit from more trials to sample the hyperparameter space and converge to the optimal values.
• Option C: Using UNIT_LOG_SCALE for the embedding dimension, UNIT_LINEAR_SCALE for the learning rate, and a large number of parallel trials, mismatches the appropriate scaling type for each hyperparameter, based on their range and distribution. The embedding dimension is an integer hyperparameter that varies linearly between 16 and 64, so using UNIT_LOG_SCALE is not suitable. The learning rate is a double hyperparameter that varies exponentially between 10e-05 and 10e-02, so using UNIT_LINEAR_SCALE makes less sense.
• Option D: Using UNIT_LOG_SCALE for the embedding dimension, UNIT_LINEAR_SCALE for the learning rate, and a small number of parallel trials, combines the drawbacks of option B and option C. It mismatches the appropriate scaling type for each hyperparameter, based on their range and distribution, and reduces the exploration of the hyperparameter space, by using a small number of parallel trials.

References:

• [Vertex AI: Hyperparameter tuning overview]
• [Vertex AI: Configuring the hyperparameter tuning job]

• Reasoning: The question asks for a design that serves asynchronous predictions to determine whether a machine part will fail. This means that the predictions do not need to be returned immediately to the sensors, but can be processed in batches and sent to a downstream system for monitoring. Option B is the only one that uses a streaming data pipeline with Pub/Sub and Dataflow, which can handle real-time data ingestion, processing, and prediction. Option B also invokes the model for prediction, which is required by the question. The other options either use synchronous predictions (option A), batch predictions (options C and D), or do not invoke the model for prediction (option D).
• References: You can learn more about the differences between synchronous, asynchronous, and batch predictions in Vertex AI from this document. You can also find examples of how to use Pub/Sub and Dataflow for streaming data pipelines from this tutorial and this codelab.

• Option A is incorrect because distributing the dataset with tf.distribute.Strategy.experimental_distribute_dataset is not the most effective way to decrease the training time. This method allows you to distribute your dataset across multiple devices or machines, by creating a tf.data.Dataset instance that can be iterated over in parallel1. However, this option may not improve the training time significantly, as it does not change the amount of data or computation that each device or machine has to process. Moreover, this option may introduce additional overhead or complexity, as it requires you to handle the data sharding, replication, and synchronization across the devices or machines1.
• Option B is incorrect because creating a custom training loop is not the easiest way to decrease the training time. A custom training loop is a way to implement your own logic for training your model, by using low-level TensorFlow APIs, such as tf.GradientTape, tf.Variable, or tf.function2. A custom training loop may give you more flexibility and control over the training process, but it also requires more effort and expertise, as you have to write and debug the code for each step of the training loop, such as computing the gradients, applying the optimizer, or updating the metrics2. Moreover, a custom training loop may not improve the training time significantly, as it does not change the amount of data or computation that each device or machine has to process.
• Option C is incorrect because using a TPU with tf.distribute.TPUStrategy is not a valid way to decrease the training time. A TPU (Tensor Processing Unit) is a custom hardware accelerator designed for high-performance ML workloads3. A tf.distribute.TPUStrategy is a distribution strategy that allows you to distribute your training across multiple TPUs, by creating a tf.distribute.TPUStrategy instance that can be used with high-level TensorFlow APIs, such as Keras4. However, this option is not feasible, as Vertex AI Training does not support TPUs as accelerators for custom training jobs5. Moreover, this option may require significant code changes, as TPUs have different requirements and limitations than GPUs.
• Option D is correct because increasing the batch size is the best way to decrease the training time. The batch size is a hyperparameter that determines how many samples of data are processed in each iteration of the training loop. Increasing the batch size may reduce the training time, as it reduces the number of iterations needed to train the model, and it allows each device or machine to process more data in parallel. Increasing the batch size is also easy to implement, as it only requires changing a single hyperparameter. However, increasing the batch size may also affect the convergence and the accuracy of the model, so it is important to find the optimal batch size that balances the trade-off between the training time and the model performance.

References:

• tf.distribute.Strategy.experimental_distribute_dataset
• Custom training loop
• TPU overview
• tf.distribute.TPUStrategy
• Vertex AI Training accelerators
• [TPU programming model]
• [Batch size and learning rate]
• [Keras overview]
• [tf.distribute.MirroredStrategy]
• [Vertex AI Training overview]
• [TensorFlow overview]

 Cloud Storage is a scalable and cost-effective storage service for any type of data. By storing image files in Cloud Storage, you can access them from anywhere and avoid the overhead of managing your own storage infrastructure. However, accessing image files directly from Cloud Storage can be slow and inefficient, especially for large-scale training. A better option is to use serialized records, such as TFRecord or Apache Avro, which are binary formats that store multiple images and their labels in a single file. Serialized records can improve the data throughput and reduce the network latency, as well as enable data compression and sharding. You can use TensorFlow or Apache Beam APIs to create and read serialized records from Cloud Storage. This solution requires minimal code changes and can speed up your training time significantly. References:

• Cloud Storage | Google Cloud
• TFRecord and tf.Example | TensorFlow Core
• Apache Avro 1.10.2 Specification
• Using Apache Beam with Cloud Storage | Cloud Storage

• Option A is incorrect because Vertex AI Pipelines and App Engine do not meet all the requirements of the system. Vertex AI Pipelines is a service that allows you to create, run, and manage ML workflows using TensorFlow Extended (TFX) components or custom components1. App Engine is a service that allows you to build and deploy scalable web applications using standard or flexible environments2. However, App Engine does not support Docker containers in the standard environment, and does not provide a dedicated service for online prediction and monitoring of ML models3.
• Option B is correct because Vertex AI Pipelines, Vertex AI Prediction, and Vertex AI Model Monitoring meet all the requirements of the system. Vertex AI Prediction is a service that allows you to deploy and serve ML models for online or batch prediction, with support for autoscaling and custom containers4. Vertex AI Model Monitoring is a service that allows you to monitor the performance and fairness of your deployed models, and get alerts for any issues or anomalies5.
• Option C is incorrect because Cloud Composer, BigQuery ML, and Vertex AI Prediction do not meet all the requirements of the system. Cloud Composer is a service that allows you to create, schedule, and manage workflows using Apache Airflow. BigQuery ML is a service that allows you to create and use ML models within BigQuery using SQL queries. However, BigQuery ML does not support custom containers, and Vertex AI Prediction does not support scheduled model retraining or model monitoring.
• Option D is incorrect because Cloud Composer, Vertex AI Training with custom containers, and App Engine do not meet all the requirements of the system. Vertex AI Training is a service that allows you to train ML models using built-in algorithms or custom containers. However, Vertex AI Training does not support online prediction or model monitoring, and App Engine does not support Docker containers in the standard environment or online prediction and monitoring of ML models3.

References:

• Vertex AI Pipelines overview
• App Engine overview
• Choosing an App Engine environment
• Vertex AI Prediction overview
• Vertex AI Model Monitoring overview
• [Cloud Composer overview]
• [BigQuery ML overview]
• [BigQuery ML limitations]
• [Vertex AI Training overview]

Vertex ML Metadata is a service that lets you track and manage the metadata produced by your ML workflows in a centralized way. It helps you have reproducible experiments by generating artifacts that represent the data, parameters, and metrics used or produced by your ML system. You can also analyze the lineage and performance of your ML artifacts using Vertex ML Metadata.

Some of the benefits of using Vertex ML Metadata are:

• It captures your ML system’s metadata as a graph, where artifacts and executions are nodes, and events are edges that link them as inputs or outputs.
• It allows you to create contexts to group sets of artifacts and executions together, such as experiments, runs, or projects.
• It supports querying and filtering the metadata using the Vertex AI SDK for Python or REST commands.
• It integrates with other Vertex AI services, such as Vertex AI Pipelines and Vertex AI Experiments, to automatically log metadata and artifacts.

The other options are not suitable for tracking and managing ML metadata in a centralized way.

• Option A: Storing your tf.logging data in BigQuery is not enough to capture the full metadata of your ML system, such as the artifacts and their lineage. BigQuery is a data warehouse service that is mainly used for analytics and reporting, not for metadata management.
• Option B: Managing all relational entities in the Hive Metastore is not a good solution for ML metadata, as it is designed for storing metadata of Hive tables and partitions, not for ML artifacts and executions. Hive Metastore is a component of the Apache Hive project, which is a data warehouse system for querying and analyzing large datasets stored in Hadoop.
• Option C: Storing all ML metadata in Google Cloud’s operations suite is not a feasible option, as it is a set of tools for monitoring, logging, tracing, and debugging your applications and infrastructure, not for ML metadata. Google Cloud’s operations suite does not provide the features and integrations that Vertex ML Metadata offers for ML workflows.

The best option for reducing pipeline execution time and cost, while also minimizing pipeline changes, is to enable caching for the pipeline job, and disable caching for the model training step. This option allows you to leverage the power and simplicity of Vertex AI Pipelines to reuse the output of the data preprocessing step, and avoid unnecessary recomputation. Vertex AI Pipelines is a service that can orchestrate machine learning workflows using Vertex AI. Vertex AI Pipelines can run preprocessing and training steps on custom Docker images, and evaluate, deploy, and monitor the machine learning model. Caching is a feature of Vertex AI Pipelines that can store and reuse the output of a pipeline step, and skip the execution of the step if the input parameters and the code have not changed. Caching can help you reduce the pipeline execution time and cost, as you do not need to re-run the same step with the same input and code. Caching can also help you minimize the pipeline changes, as you do not need to add or remove any pipeline steps or parameters. By enabling caching for the pipeline job, and disabling caching for the model training step, you can create a Vertex AI pipeline that includes two steps. The first step preprocesses 10 TB data, completes in about 1 hour, and saves the result in a Cloud Storage bucket. The second step uses the processed data to train a model. You can update the model’s code to allow you to test different algorithms, and run the pipeline job with caching enabled. The pipeline job will reuse the output of the data preprocessing step from the cache, and skip the execution of the step. The pipeline job will run the model training step with the updated code, and disable the caching for the step. This way, you can reduce the pipeline execution time and cost, while also minimizing pipeline changes1.

The other options are not as good as option D, for the following reasons:

• Option A: Adding a pipeline parameter and an additional pipeline step, depending on the parameter value, the pipeline step conducts or skips data preprocessing and starts model training, would require more skills and steps than enabling caching for the pipeline job, and disabling caching for the model training step. A pipeline parameter is a variable that can be used to control the input or output of a pipeline step. A pipeline parameter can help you customize the pipeline logic and behavior, and experiment with different values. An additional pipeline step is a new instance of a pipeline component that can perform a part of the pipeline workflow, such as data preprocessing or model training. An additional pipeline step can help you extend the pipeline functionality and complexity, and handle different scenarios. However, adding a pipeline parameter and an additional pipeline step, depending on the parameter value, the pipeline step conducts or skips data preprocessing and starts model training, would require more skills and steps than enabling caching for the pipeline job, and disabling caching for the model training step. You would need to write code, define the pipeline parameter, create the additional pipeline step, implement the conditional logic, and compile and run the pipeline. Moreover, this option would not reuse the output of the data preprocessing step from the cache, but rather from the Cloud Storage bucket, which can increase the data transfer and access costs1.
• Option B: Creating another pipeline without the preprocessing step, and hardcoding the preprocessed Cloud Storage file location for model training, would require more skills and steps than enabling caching for the pipeline job, and disabling caching for the model training step. A pipeline without the preprocessing step is a pipeline that only includes the model training step, and uses the preprocessed data from the Cloud Storage bucket as the input. A pipeline without the preprocessing step can help you avoid running the data preprocessing step every time, and reduce the pipeline execution time and cost. However, creating another pipeline without the preprocessing step, and hardcoding the preprocessed Cloud Storage file location for model training, would require more skills and steps than enabling caching for the pipeline job, and disabling caching for the model training step. You would need to write code, create a new pipeline, remove the preprocessing step, hardcode the Cloud Storage file location, and compile and run the pipeline. Moreover, this option would not reuse the output of the data preprocessing step from the cache, but rather from the Cloud Storage bucket, which can increase the data transfer and access costs. Furthermore, this option would create another pipeline, which can increase the maintenance and management costs1.
• Option C: Configuring a machine with more CPU and RAM from the compute-optimized machine family for the data preprocessing step, would not reduce the pipeline execution time and cost, while also minimizing pipeline changes, but rather increase the pipeline execution cost and complexity. A machine with more CPU and RAM from the compute-optimized machine family is a virtual machine that has a high ratio of CPU cores to memory, and can provide high performance and scalability for compute-intensive workloads. A machine with more CPU and RAM from the compute-optimized machine family can help you optimize the data preprocessing step, and reduce the pipeline execution time. However, configuring a machine with more CPU and RAM from the compute-optimized machine family for the data preprocessing step, would not reduce the pipeline execution time and cost, while also minimizing pipeline changes, but rather increase the pipeline execution cost and complexity. You would need to write code, configure the machine type parameters for the data preprocessing step, and compile and run the pipeline. Moreover, this option would increase the pipeline execution cost, as machines with more CPU and RAM from the compute-optimized machine family are more expensive than machines with less CPU and RAM from other machine families. Furthermore, this option would not reuse the output of the data preprocessing step from the cache, but rather re-run the data preprocessing step every time, which can increase the pipeline execution time and cost1.

References:

• Preparing for Google Cloud Certification: Machine Learning Engineer, Course 3: Production ML Systems, Week 3: MLOps
• Google Cloud Professional Machine Learning Engineer Exam Guide, Section 3: Scaling ML models in production, 3.2 Automating ML workflows
• Official Google Cloud Certified Professional Machine Learning Engineer Study Guide, Chapter 6: Production ML Systems, Section 6.4: Automating ML Workflows
• Vertex AI Pipelines
• Caching
• Pipeline parameters
• Machine types

According to the official exam guide1, one of the skills assessed in the exam is to “configure and optimize model monitoring jobs”. The Vertex AI Model Monitoring documentation states that “to reduce the cost of model monitoring, you can configure the sample rate of the requests that are logged and analyzed by model monitoring”. Therefore, decreasing the sample_rate parameter in the Randomsampleconfig of the monitoring job would reduce the model monitoring cost while continuing to quickly detect drift. The other options are not relevant or optimal for this scenario. References:

• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate
• Professional ML Engineer Exam Guide
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions
• [Vertex AI Model Monitoring]

 A predict request is a way to send data to a trained model and get predictions in return. A predict request can be written in different formats, such as JSON, protobuf, or gRPC, depending on the service and the platform that are used to host and serve the model. A predict request usually contains the following information:

• The signature name: This is the name of the signature that defines the inputs and outputs of the model. A signature is a way to specify the expected format, type, and shape of the data that the model can accept and produce. A signature can be specified when exporting or saving the model, or it can be automatically inferred by the service or the platform. A model can have multiple signatures, but only one can be used for each predict request.
• The instances: This is the data that is sent to the model for prediction. The instances can be a single instance or a batch of instances, depending on the size and shape of the data. The instances should match the input specification of the signature, such as the number, name, and type of the input tensors.

For the use case of training a text classification model, the correct way to write the predict request is D. data = json.dumps({“signature_name”: “serving_default”, “instances”: [[‘a’, ‘b’], [‘c’, ‘d’], [‘e’, ‘f’]]})

This option involves writing the predict request in JSON format, which is a common and convenient format for sending and receiving data over the web. JSON stands for JavaScript Object Notation, and it is a way to represent data as a collection of name-value pairs or an ordered list of values. JSON can be easily converted to and from Python objects using the json module.

This option also involves using the signature name “serving_default”, which is the default signature name that is assigned to the model when it is saved or exported without specifying a custom signature name. The serving_default signature defines the input and output tensors of the model based on the SignatureDef that is shown in the image. According to the SignatureDef, the model expects an input tensor called “text” that has a shape of (-1, 2) and a type of DT_STRING, and produces an output tensor called “softmax” that has a shape of (-1, 2) and a type of DT_FLOAT. The -1 in the shape indicates that the dimension can vary depending on the number of instances, and the 2 indicates that the dimension is fixed at 2. The DT_STRING and DT_FLOAT indicate that the data type is string and float, respectively.

This option also involves sending a batch of three instances to the model for prediction. Each instance is a list of two strings, such as [‘a’, ‘b’], [‘c’, ‘d’], or [‘e’, ‘f’]. These instances match the input specification of the signature, as they have a shape of (3, 2) and a type of string. The model will process these instances and produce a batch of three predictions, each with a softmax output that has a shape of (1, 2) and a type of float. The softmax output is a probability distribution over the two possible classes that the model can predict, such as positive or negative sentiment.

Therefore, writing the predict request as data = json.dumps({“signature_name”: “serving_default”, “instances”: [[‘a’, ‘b’], [‘c’, ‘d’], [‘e’, ‘f’]]}) is the correct and valid way to send data to the text classification model and get predictions in return.

References:

• [json — JSON encoder and decoder]

The class imbalance problem is a common challenge in machine learning, especially in classification tasks. It occurs when the distribution of the target classes is highly skewed, such that one class (the majority class) has much more examples than the other class (the minority class). The minority class is often the more interesting or important class, such as failure incidents, fraud cases, or rare diseases. However, most machine learning algorithms are designed to optimize the overall accuracy, which can be biased towards the majority class and ignore the minority class. This can result in poor predictive performance, especially for the minority class.

There are different techniques to deal with the class imbalance problem, such as data-level methods, algorithm-level methods, and evaluation-level methods1. Data-level methods involve resampling the original dataset to create a more balanced class distribution. There are two main types of data-level methods: oversampling and undersampling. Oversampling methods increase the number of examples in the minority class, either by duplicating existing examples or by generating synthetic examples. Undersampling methods reduce the number of examples in the majority class, either by randomly removing examples or by using clustering or other criteria to select representative examples. Both oversampling and undersampling methods can be combined with upweighting or downweighting, which assign different weights to the examples according to their class frequency, to further balance the dataset.

For the use case of investigating failures of a production line component based on sensor readings, the best option is to downsample the data with upweighting to create a sample with 10% positive examples. This option involves randomly removing some of the negative examples (the majority class) until the ratio of positive to negative examples is 1:9, and then assigning higher weights to the positive examples to compensate for their low frequency. This option can create a more balanced dataset that can improve the performance of the classification models, while preserving the diversity and representativeness of the original data. This option can also reduce the computation time and memory usage, as the size of the dataset is reduced. Therefore, downsampling the data with upweighting to create a sample with 10% positive examples is the best option for this use case.

References:

• A Systematic Study of the Class Imbalance Problem in Convolutional Neural Networks

The best option for scaling a Vertex AI endpoint efficiently when the demand increases in the future, using a scikit-learn model that is deployed to a Vertex AI endpoint and tested on live production traffic, is to configure an appropriate minReplicaCount value based on expected baseline traffic. This option allows you to leverage the power and simplicity of Vertex AI to automatically scale your endpoint resources according to the traffic patterns. Vertex AI is a unified platform for building and deploying machine learning solutions on Google Cloud. Vertex AI can deploy a trained model to an online prediction endpoint, which can provide low-latency predictions for individual instances. Vertex AI can also provide various tools and services for data analysis, model development, model deployment, model monitoring, and model governance. A minReplicaCount value is a parameter that specifies the minimum number of replicas that the endpoint must always have, regardless of the load. A minReplicaCount value can help you ensure that the endpoint has enough resources to handle the expected baseline traffic, and avoid high latency or errors. By configuring an appropriate minReplicaCount value based on expected baseline traffic, you can scale your endpoint efficiently when the demand increases in the future. You can set the minReplicaCount value when you deploy the model to the endpoint, or update it later. Vertex AI will automatically scale up or down the number of replicas within the range of the minReplicaCount and maxReplicaCount values, based on the target utilization percentage and the autoscaling metric1.

The other options are not as good as option B, for the following reasons:

• Option A: Deploying two models to the same endpoint and distributing requests among them evenly would not allow you to scale your endpoint efficiently when the demand increases in the future, and could increase the complexity and cost of the deployment process. A model is a resource that represents a machine learning model that you can use for prediction. A model can have one or more versions, which are different implementations of the same model. A model version can help you experiment and iterate on your model, and improve the model performance and accuracy. An endpoint is a resource that provides the service endpoint (URL) you use to request the prediction. An endpoint can have one or more deployed models, which are instances of model versions that are associated with physical resources. A deployed model can help you serve online predictions with low latency, and scale up or down based on the traffic. By deploying two models to the same endpoint and distributing requests among them evenly, you can create a load balancing mechanism that can distribute the traffic across the models, and reduce the load on each model. However, deploying two models to the same endpoint and distributing requests among them evenly would not allow you to scale your endpoint efficiently when the demand increases in the future, and could increase the complexity and cost of the deployment process. You would need to write code, create and configure the two models, deploy the models to the same endpoint, and distribute the requests among them evenly. Moreover, this option would not use the autoscaling feature of Vertex AI, which can automatically adjust the number of replicas based on the traffic patterns, and provide various benefits, such as optimal resource utilization, cost savings, and performance improvement2.
• Option C: Setting the target utilization percentage in the autoscalingMetricSpecs configuration to a higher value would not allow you to scale your endpoint efficiently when the demand increases in the future, and could cause errors or poor performance. A target utilization percentage is a parameter that specifies the desired utilization level of each replica. A target utilization percentage can affect the speed and accuracy of the autoscaling process. A higher target utilization percentage can help you reduce the number of replicas, but it can also cause high latency, low throughput, or resource exhaustion. By setting the target utilization percentage in the autoscalingMetricSpecs configuration to a higher value, you can increase the utilization level of each replica, and save some resources. However, setting the target utilization percentage in the autoscalingMetricSpecs configuration to a higher value would not allow you to scale your endpoint efficiently when the demand increases in the future, and could cause errors or poor performance. You would need to write code, create and configure the autoscalingMetricSpecs, and set the target utilization percentage to a higher value. Moreover, this option would not ensure that the endpoint has enough resources to handle the expected baseline traffic, which could cause high latency or errors1.
• Option D: Changing the model’s machine type to one that utilizes GPUs would not allow you to scale your endpoint efficiently when the demand increases in the future, and could increase the complexity and cost of the deployment process. A machine type is a parameter that specifies the type of virtual machine that the prediction service uses for the deployed model. A machine type can affect the speed and accuracy of the prediction process. A machine type that utilizes GPUs can help you accelerate the computation and processing of the prediction, and handle more prediction requests at the same time. By changing the model’s machine type to one that utilizes GPUs, you can improve the prediction performance and efficiency of your model. However, changing the model’s machine type to one that utilizes GPUs would not allow you to scale your endpoint efficiently when the demand increases in the future, and could increase the complexity and cost of the deployment process. You would need to write code, create and configure the model, deploy the model to the endpoint, and change the machine type to one that utilizes GPUs. Moreover, this option would not use the autoscaling feature of Vertex AI, which can automatically adjust the number of replicas based on the traffic patterns, and provide various benefits, such as optimal resource utilization, cost savings, and performance improvement2.

References:

• Configure compute resources for prediction | Vertex AI | Google Cloud
• Deploy a model to an endpoint | Vertex AI | Google Cloud

 BigQuery is a service that allows you to store and query large amounts of data in a scalable and cost-effective way. You can use BigQuery to build a model that predicts customer lifetime value over the next three years, by using the create model statement. The create model statement is a SQL command that allows you to create and train an ML model using your data in BigQuery. You can use the create model statement to create an AutoML model, which is a type of model that automatically selects the best features and architecture for your data. By using an AutoML model, you can use the simplest approach to build the model, without writing any code or performing any feature engineering. You can also use the ml.evaluate and ml.predict statements to validate the results of your model. The ml.evaluate statement is a SQL command that allows you to evaluate the performance and quality of your model using various metrics. The ml.predict statement is a SQL command that allows you to make predictions using your model and new data. You can also use the BigQuery console to access visualization tools, such as charts and graphs, to explore and analyze your data and model results. By using the BigQuery console, the create model statement, and the ml.evaluate and ml.predict statements, you can build and validate a model that predicts customer lifetime value over the next three years, and have access to visualization tools. References:

• BigQuery documentation
• create model statement documentation
• ml.evaluate statement documentation
• ml.predict statement documentation
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

Developing ML models with AI Platform for image segmentation on CT scans requires a lot of computation and experimentation, as image segmentation is a complex and challenging task that involves assigning a label to each pixel in an image. Image segmentation can be used for various medical applications, such as tumor detection, organ segmentation, or lesion localization1

To minimize the computation costs and manual intervention while having version control for the code, one should use Cloud Build linked with Cloud Source Repositories to trigger retraining when new code is pushed to the repository. Cloud Build is a service that executes your builds on Google Cloud Platform infrastructure. Cloud Build can import source code from Cloud Source Repositories, Cloud Storage, GitHub, or Bitbucket, execute a build to your specifications, and produce artifacts such as Docker containers or Java archives2

Cloud Build allows you to set up automated triggers that start a build when changes are pushed to a source code repository. You can configure triggers to filter the changes based on the branch, tag, or file path3

Cloud Source Repositories is a service that provides fully managed private Git repositories on Google Cloud Platform. Cloud Source Repositories allows you to store, manage, and track your code using the Git version control system. You can also use Cloud Source Repositories to connect to other Google Cloud services, such as Cloud Build, Cloud Functions, or Cloud Run4

To use Cloud Build linked with Cloud Source Repositories to trigger retraining when new code is pushed to the repository, you need to do the following steps:

• Create a Cloud Source Repository for your code, and push your code to the repository. You can use the Cloud SDK, Cloud Console, or Cloud Source Repositories API to create and manage your repository5
• Create a Cloud Build trigger for your repository, and specify the build configuration and the trigger settings. You can use the Cloud SDK, Cloud Console, or Cloud Build API to create and manage your trigger.
• Specify the steps of the build in a YAML or JSON file, such as installing the dependencies, running the tests, building the container image, and submitting the training job to AI Platform. You can also use the Cloud Build predefined or custom build steps to simplify your build configuration.
• Push your new code to the repository, and the trigger will start the build automatically. You can monitor the status and logs of the build using the Cloud SDK, Cloud Console, or Cloud Build API.

The other options are not as easy or feasible. Using Cloud Functions to identify changes to your code in Cloud Storage and trigger a retraining job is not ideal, as Cloud Functions has limitations on the memory, CPU, and execution time, and does not provide a user interface for managing and tracking your builds. Using the gcloud command-line tool to submit training jobs on AI Platform when you update your code is not optimal, as it requires manual intervention and does not leverage the benefits of Cloud Build and its integration with Cloud Source Repositories. Creating an automated workflow in Cloud Composer that runs daily and looks for changes in code in Cloud Storage using a sensor is not relevant, as Cloud Composer is mainly designed for orchestrating complex workflows across multiple systems, and does not provide a version control system for your code.

References: 1: Image segmentation 2: Cloud Build overview 3: Creating and managing build triggers 4: Cloud Source Repositories overview 5: Quickstart: Create a repository : [Quickstart: Create a build trigger] : [Configuring builds] : [Viewing build results]

The option A is the most suitable solution for logging data and tracking artifacts from each run of a model development experiment in a Vertex AI Workbench notebook. Vertex AI Workbench is a service that allows you to create and run interactive notebooks on Google Cloud. You can use Vertex AI Workbench to experiment with different preprocessing and modeling approaches for your time series prediction problem. You can also use the Vertex AI TensorBoard instance and the Vertex AI SDK to create an experiment and associate the TensorBoard instance. TensorBoard is a tool that allows you to visualize and monitor the metrics and artifacts of your ML experiments. You can use the Vertex AI SDK to create an experiment object, which is a logical grouping of runs that share a common objective. You can also use the Vertex AI SDK to associate the experiment object with a TensorBoard instance, which is a managed service that hosts a TensorBoard web app. By using the Vertex AI TensorBoard instance and the Vertex AI SDK, you can easily set up and manage your experiments, and access the TensorBoard web app from the Vertex AI console. You can also use the log_time_series_metrics function and the log_metrics function to log data and track artifacts from each run. The log_time_series_metrics function is a function that allows you to log the time series data, such as the multivariate time series and the labels, to the TensorBoard instance. The log_metrics function is a function that allows you to log the scalar metrics, such as the loss values, to the TensorBoard instance. By using these functions, you can record the data and artifacts from each run of your experiment, and compare them in the TensorBoard web app. You can also use the TensorBoard web app to visualize the data and artifacts, such as the time series plots, the scalar charts, the histograms, and the distributions. By using the Vertex AI TensorBoard instance, the Vertex AI SDK, and the log functions, you can log data and track artifacts from each run of your experiment in a Vertex AI Workbench notebook. References:

• Vertex AI Workbench documentation
• Vertex AI TensorBoard documentation
• Vertex AI SDK documentation
• log_time_series_metrics function documentation
• log_metrics function documentation
• [Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate]

According to the official exam guide1, one of the skills assessed in the exam is to “design, build, and productionalize ML models to solve business challenges using Google Cloud technologies”. Vertex AI AutoML Forecasting2 is a service that allows you to train and deploy custom time-series forecasting models for batch prediction. Vertex AI AutoML Forecasting simplifies the model development process by providing a graphical user interface and a no-code approach. You can use Vertex AI AutoML Forecasting to train a model by using your tabular data, and specify the target variable, the covariates, and the time variable. Vertex AI AutoML Forecasting automatically handles the feature engineering, model selection, and hyperparameter tuning. Therefore, option D is the best way to maximize the speed of model development and testing for the given use case. The other options are not relevant or optimal for this scenario. References:

• Professional ML Engineer Exam Guide
• Vertex AI AutoML Forecasting
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

 This answer is correct because it uses Firebase, a platform that provides a scalable and reliable notification system for mobile and web applications. Firebase Cloud Messaging (FCM) allows you to send messages and notifications to users across different devices and platforms. By registering each user with a user ID on the FCM server, you can target specific users based on their account balance predictions and send them personalized notifications when their balance is likely to drop below the $25 threshold. This way, you can provide a useful and timely feature for your customers and increase their engagement and retention. References:

• [Firebase Cloud Messaging]
• [Firebase Cloud Messaging: Send messages to specific devices]

To avoid creating or reinforcing unfair bias in the model, you should collect a representative sample of production traffic to build the training dataset, and conduct fairness tests across sensitive categories and demographics on the trained model. A representative sample is one that reflects the true distribution of the population, and does not over- or under-represent any group. A random sample is a simple way to obtain a representative sample, as it ensures that every data point has an equal chance of being selected. A stratified sample is another way to obtain a representative sample, as it ensures that every subgroup has a proportional representation in the sample. However, a stratified sample requires prior knowledge of the subgroups and their sizes, which may not be available or easy to obtain. Therefore, a random sample is a more feasible option in this case. A fairness test is a way to measure and evaluate the potential bias and discrimination of the model, based on different categories and demographics, such as age, gender, race, etc. A fairness test can help you identify and mitigate any unfair outcomes or impacts of the model, and ensure that the model treats all groups fairly and equitably. A fairness test can be conducted using various methods and tools, such as confusion matrices, ROC curves, fairness indicators, etc. References: The answer can be verified from official Google Cloud documentation and resources related to data sampling and fairness testing.

• Sampling data | BigQuery
• Fairness Indicators | TensorFlow
• What-if Tool | TensorFlow

To log the metrics of a machine learning model in TensorFlow using the Vertex AI Python SDK, you should utilize the aiplatform.log_metrics function to log the F1 score and aiplatform.log_classification_metrics function to log the confusion matrix. These functions allow users to manually record and store evaluation metrics for each model, facilitating an efficient comparison based on specific performance indicators like F1 scores and confusion matrices. References: The answer can be verified from official Google Cloud documentation and resources related to Vertex AI and TensorFlow.

• Vertex AI Python SDK reference | Google Cloud
• Logging custom metrics | Vertex AI
• Migrating from scikit-learn to TensorFlow | TensorFlow

The best option for developing and comparing multiple models on a large-scale BigQuery table using TensorFlow and Vertex AI is to use TensorFlow I/O’s BigQuery Reader to directly read the data. This option has the following advantages:

• It minimizes any bottlenecks during the data ingestion stage, as the BigQuery Reader can stream data from BigQuery to TensorFlow in parallel and in batches, without loading the entire table into memory or disk. The BigQuery Reader can also perform data transformations and filtering using SQL queries, reducing the need for additional preprocessing steps in TensorFlow.
• It leverages the scalability and performance of BigQuery, as the BigQuery Reader can handle hundreds of millions of records worth of training data efficiently and reliably. BigQuery is a serverless, fully managed, and highly scalable data warehouse that can run complex queries over petabytes of data in seconds.
• It simplifies the integration with Vertex AI, as the BigQuery Reader can be used with both custom and pre-built TensorFlow models on Vertex AI. Vertex AI is a unified platform for machine learning that provides various tools and features for data ingestion, data labeling, data preprocessing, model training, model tuning, model deployment, model monitoring, and model explainability.

The other options are less optimal for the following reasons:

• Option A: Using the BigQuery client library to load data into a dataframe, and using tf.data.Dataset.from_tensor_slices() to read it, introduces memory and performance issues. This option requires loading the entire BigQuery table into a Pandas dataframe, which can consume a lot of memory and cause out-of-memory errors. Moreover, using tf.data.Dataset.from_tensor_slices() to read the dataframe can be slow and inefficient, as it creates one slice per row of the dataframe, resulting in a large number of small tensors.
• Option B: Exporting data to CSV files in Cloud Storage, and using tf.data.TextLineDataset() to read them, introduces additional steps and complexity. This option requires exporting the BigQuery table to one or more CSV files in Cloud Storage, which can take a long time and consume a lot of storage space. Moreover, using tf.data.TextLineDataset() to read the CSV files can be slow and error-prone, as it requires parsing and decoding each line of text, handling missing values and invalid data, and applying data transformations and validations.
• Option C: Converting the data into TFRecords, and using tf.data.TFRecordDataset() to read them, introduces additional steps and complexity. This option requires converting the BigQuery table into one or more TFRecord files, which are binary files that store serialized TensorFlow examples. This can take a long time and consume a lot of storage space. Moreover, using tf.data.TFRecordDataset() to read the TFRecord files requires defining and parsing the schema of the TensorFlow examples, which can be tedious and error-prone.

References:

• [TensorFlow I/O documentation]
• [BigQuery documentation]
• [Vertex AI documentation]

TensorFlow Serving is a service that allows you to deploy and serve TensorFlow models in a scalable and efficient way. TensorFlow Serving supports various platforms and hardware, such as CPU, GPU, and TPU. However, the default TensorFlow Serving binaries are built with generic CPU instructions, which may not leverage the full potential of the CPU architecture. To improve the serving latency and performance, you can recompile TensorFlow Serving using the source code and enable CPU-specific optimizations, such as AVX, AVX2, and FMA1. These optimizations can speed up the computation and inference of the TensorFlow models, especially for deep neural networks.

Google Kubernetes Engine (GKE) is a service that allows you to run and manage containerized applications on Google Cloud using Kubernetes. GKE supports various types and sizes of nodes, which are the virtual machines that run the containers. GKE also supports different CPU platforms, which are the generations and models of the CPUs that power the nodes. GKE allows you to choose a baseline minimum CPU platform for your node pool, which is a group of nodes with the same configuration. By choosing a baseline minimum CPU platform, you can ensure that your nodes have the CPU features and capabilities that match your workload requirements2.

For the use case of serving a few thousand queries per second and experiencing latency issues, the best option is to recompile TensorFlow Serving using the source to support CPU-specific optimizations, and instruct GKE to choose an appropriate baseline minimum CPU platform for serving nodes. This option can improve the serving latency and performance without changing the underlying infrastructure, as it only involves rebuilding the TensorFlow Serving binary and selecting the CPU platform for the GKE nodes. This option can also take advantage of the CPU-only pods that are running on GKE, as it can optimize the CPU utilization and efficiency. Therefore, recompiling TensorFlow Serving using the source to support CPU-specific optimizations and instructing GKE to choose an appropriate baseline minimum CPU platform for serving nodes is the best option for this use case.

References:

• Building TensorFlow Serving from source
• Specifying a minimum CPU platform for a node pool

The number of trainable weights in a DNN regression model with Keras APIs can be calculated by multiplying the number of input units by the number of output units for each layer, and adding the number of bias units for each layer. The bias units are usually equal to the number of output units, except for the last layer, which does not have bias units if the activation function is softmax1. In this code, the model has three layers: a dense layer with 256 units and relu activation, a dropout layer with 0.25 rate, and a dense layer with 2 units and softmax activation. The input shape is 500. Therefore, the number of trainable weights is:

• For the first layer: 500 input units * 256 output units + 256 bias units = 128256
• For the second layer: The dropout layer does not have any trainable weights, as it only randomly sets some of the input units to zero to prevent overfitting2.
• For the third layer: 256 input units * 2 output units + 0 bias units = 512

The total number of trainable weights is 128256 + 512 = 161024. Therefore, the correct answer is B.

References:

• How to calculate the number of parameters for a Convolutional Neural Network?
• Dropout (keras.io)

AutoML Translation is a service that allows you to create and train custom ML models for translating text between different languages. You can use AutoML Translation to train a model that can translate instruction manuals for scientific products to 15 different languages. You can also use Translation Hub to configure a project and use the trained model to translate the documents. Translation Hub is a service that allows you to manage and automate your translation workflows on Google Cloud. You can use Translation Hub to upload the documents to a Cloud Storage bucket, select the source and target languages, and apply the trained model to translate the documents. You can also use Translation Hub to download the translated documents or save them to another Cloud Storage bucket. You can also use human reviewers to evaluate the incorrect translations. Human reviewers are people who can review and correct the translations produced by the ML model. You can use human reviewers to improve the quality and accuracy of the translations, and provide feedback to the ML model. You can use Translation Hub to integrate with third-party human review services, such as Google Translate Community or Appen. By using AutoML Translation, Translation Hub, and human reviewers, you can implement a scalable solution that maximizes accuracy and minimizes operational overhead. You can also include a process to evaluate and fix incorrect translations. References:

• [AutoML Translation documentation]
• [Translation Hub documentation]
• [Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate]

AutoML Natural Language is a service that allows you to quickly build, test and deploy natural language processing (NLP) models without needing to have expertise in NLP or machine learning. You can use it to train a classifier on your corpus of written support cases, and then use the AutoML API to perform classification on new requests. Once the model is trained, it can be deployed as a REST API. This allows the classifier to be integrated into your pipeline and be easily consumed by other systems.

According to the official exam guide1, one of the skills assessed in the exam is to “automate and orchestrate ML pipelines using Cloud Composer”. Vertex AI Pipelines2 is a service that allows you to orchestrate your ML workflows using Kubeflow Pipelines SDK v2 or TensorFlow Extended. Vertex AI Pipelines supports execution caching, which means that if you run a pipeline and it reaches a component that has already been run with the same inputs and parameters, the component does not run again. Instead, the component uses the output from the previous run. This can save you time and resources when you are iterating on your pipeline. Therefore, option A is the best way to reduce model development costs, as it enables execution caching for the data export and preprocessing steps, which are likely to be the same for each model iteration. The other options are not relevant or optimal for this scenario. References:

• Professional ML Engineer Exam Guide
• Vertex AI Pipelines
• Google Professional Machine Learning Certification Exam 2023
• Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

The best option for deploying a scikit-learn classification model to production is to deploy an online Vertex AI prediction endpoint and set the max replica count to 100. This option allows you to leverage the power and scalability of Google Cloud to serve requests 24/7 and handle millions of requests per second. Vertex AI is a unified platform for building and deploying machine learning solutions on Google Cloud. Vertex AI can deploy a trained scikit-learn model to an online prediction endpoint, which can provide low-latency predictions for individual instances. An online prediction endpoint consists of one or more replicas, which are copies of the model that run on virtual machines. The max replica count is a parameter that determines the maximum number of replicas that can be created for the endpoint. By setting the max replica count to 100, you can enable the endpoint to scale up to 100 replicas when the traffic increases, and scale down to zero replicas when the traffic decreases. This can help minimize the cost of deployment, as you only pay for the resources that you use. Moreover, you can use the autoscaling algorithm option to optimize the scaling behavior of the endpoint based on the latency and utilization metrics1.

The other options are not as good as option B, for the following reasons:

• Option A: Deploying an online Vertex AI prediction endpoint and setting the max replica count to 1 would not be able to serve requests 24/7 and handle millions of requests per second. Setting the max replica count to 1 would limit the endpoint to only one replica, which can cause performance issues and service disruptions when the traffic increases. Moreover, setting the max replica count to 1 would prevent the endpoint from scaling down to zero replicas when the traffic decreases, which can increase the cost of deployment, as you pay for the resources that you do not use1.
• Option C: Deploying an online Vertex AI prediction endpoint with one GPU per replica and setting the max replica count to 1 would not be able to serve requests 24/7 and handle millions of requests per second, and would increase the cost of deployment. Adding a GPU to each replica would increase the computational power of the endpoint, but it would also increase the cost of deployment, as GPUs are more expensive than CPUs. Moreover, setting the max replica count to 1 would limit the endpoint to only one replica, which can cause performance issues and service disruptions when the traffic increases, and prevent the endpoint from scaling down to zero replicas when the traffic decreases1. Furthermore, scikit-learn models do not benefit from GPUs, as scikit-learn is not optimized for GPU acceleration2.
• Option D: Deploying an online Vertex AI prediction endpoint with one GPU per replica and setting the max replica count to 100 would be able to serve requests 24/7 and handle millions of requests per second, but it would increase the cost of deployment. Adding a GPU to each replica would increase the computational power of the endpoint, but it would also increase the cost of deployment, as GPUs are more expensive than CPUs. Setting the max replica count to 100 would enable the endpoint to scale up to 100 replicas when the traffic increases, and scale down to zero replicas when the traffic decreases, which can help minimize the cost of deployment. However, scikit-learn models do not benefit from GPUs, as scikit-learn is not optimized for GPU acceleration2. Therefore, using GPUs for scikit-learn models would be unnecessary and wasteful.

References:

• Preparing for Google Cloud Certification: Machine Learning Engineer, Course 3: Production ML Systems, Week 2: Serving ML Predictions
• Google Cloud Professional Machine Learning Engineer Exam Guide, Section 3: Scaling ML models in production, 3.1 Deploying ML models to production
• Official Google Cloud Certified Professional Machine Learning Engineer Study Guide, Chapter 6: Production ML Systems, Section 6.2: Serving ML Predictions
• Online prediction
• Scaling online prediction
• scikit-learn FAQ

 The input/output execution performance of a TensorFlow model depends on how efficiently the model can read and process the data from the data source. Reading and processing data from CSV files can be slow and inefficient, especially if the data is large and distributed. Therefore, to improve the input/output execution performance, one should use a more suitable data format and storage system.

One of the best options for improving the input/output execution performance is to convert the CSV files into shards of TFRecords, and store the data in Cloud Storage. TFRecord is a binary data format that can store a sequence of serialized TensorFlow examples. TFRecord has several advantages over CSV, such as:

• Faster data loading: TFRecord can be read and processed faster than CSV, as it avoids the overhead of parsing and decoding the text data. TFRecord also supports compression and checksums, which can reduce the data size and ensure data integrity1
• Better performance: TFRecord can improve the performance of the model, as it allows the model to access the data in a sequential and streaming manner, and leverage the tf.data API to build efficient data pipelines. TFRecord also supports sharding and interleaving, which can increase the parallelism and throughput of the data processing2
• Easier integration: TFRecord can integrate seamlessly with TensorFlow, as it is the native data format for TensorFlow. TFRecord also supports various types of data, such as images, text, audio, and video, and can store the data schema and metadata along with the data3

Cloud Storage is a scalable and reliable object storage service that can store any amount of data. Cloud Storage has several advantages over other storage systems, such as:

• High availability: Cloud Storage can provide high availability and durability for the data, as it replicates the data across multiple regions and zones, and supports versioning and lifecycle management. Cloud Storage also offers various storage classes, such as Standard, Nearline, Coldline, and Archive, to meet different performance and cost requirements4
• Low latency: Cloud Storage can provide low latency and high bandwidth for the data, as it supports HTTP and HTTPS protocols, and integrates with other Google Cloud services, such as AI Platform, Dataflow, and BigQuery. Cloud Storage also supports resumable uploads and downloads, and parallel composite uploads, which can improve the data transfer speed and reliability5
• Easy access: Cloud Storage can provide easy access and management for the data, as it supports various tools and libraries, such as gsutil, Cloud Console, and Cloud Storage Client Libraries. Cloud Storage also supports fine-grained access control and encryption, which can ensure the data security and privacy.

The other options are not as effective or feasible. Loading the data into BigQuery and reading the data from BigQuery is not recommended, as BigQuery is mainly designed for analytical queries on large-scale data, and does not support streaming or real-time data processing. Loading the data into Cloud Bigtable and reading the data from Bigtable is not ideal, as Cloud Bigtable is mainly designed for low-latency and high-throughput key-value operations on sparse and wide tables, and does not support complex data types or schemas. Converting the CSV files into shards of TFRecords and storing the data in the Hadoop Distributed File System (HDFS) is not optimal, as HDFS is not natively supported by TensorFlow, and requires additional configuration and dependencies, such as Hadoop, Spark, or Beam.

References: 1: TFRecord and tf.Example 2: Better performance with the tf.data API 3: TensorFlow Data Validation 4: Cloud Storage overview 5: Performance : [How-to guides]

• Dataflow is a fully managed service for executing Apache Beam pipelines that can process streaming or batch data1.
• Al Platform is a unified platform that enables you to build and run machine learning applications across Google Cloud2.
• BigQuery is a serverless, highly scalable, and cost-effective cloud data warehouse designed for business agility3.

These services are suitable for building an ML model to detect anomalies in real-time sensor data, as they can handle large-scale data ingestion, preprocessing, training, serving, storage, and visualization. The other options are not as suitable because:

• DataProc is a service for running Apache Spark and Apache Hadoop clusters, which are not optimized for streaming data processing4.
• AutoML is a suite of machine learning products that enables developers with limited machine learning expertise to train high-quality models specific to their business needs5. However, it does not support custom models or real-time predictions.
• Cloud Bigtable is a scalable, fully managed NoSQL database service for large analytical and operational workloads. However, it is not designed for ad hoc queries or interactive analysis.
• Cloud Functions is a serverless execution environment for building and connecting cloud services. However, it is not suitable for storing or visualizing data.
• Cloud Storage is a service for storing and accessing data on Google Cloud. However, it is not a data warehouse and does not support SQL queries or visualization tools.

The most likely reason for the observed result is that the model is overfitting in areas with less traffic and underfitting in areas with more traffic. Overfitting means that the model learns the specific patterns and noise in the training data, but fails to generalize well to new and unseen data. Underfitting means that the model is not able to capture the complexity and variability of the data, and performs poorly on both training and test data. In this case, the model might have learned to segment the images well when there is less traffic, but it might not have enough data or features to handle the more challenging scenarios when there is more traffic. This could lead to a decrease in the AUC metric, which measures the ability of the model to distinguish between different classes. AUC is a suitable metric for this classification model, as it is not affected by class imbalance or threshold selection. The other options are not likely to be the reason for the result, as they are not related to the traffic density. Too much data representing congested areas would not cause the model to fail in those areas, but rather help the model learn better. Gradients vanishing or exploding is a problem that occurs during the training process, not after the deployment, and it affects the whole model, not specific scenarios. References:

• Image Segmentation: U-Net For Self Driving Cars
• Intelligent Semantic Segmentation for Self-Driving Vehicles Using Deep Learning
• Sharing Pixelopolis, a self-driving car demo from Google I/O built with TensorFlow Lite
• Google Cloud launches machine learning engineer certification
• Google Professional Machine Learning Engineer Certification
• Professional ML Engineer Exam Guide
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

AutoML Vision Edge is a service that allows you to create custom image classification and object detection models that can run on edge devices, such as mobile phones, tablets, or IoT devices1. AutoML Vision Edge offers four types of models that vary in size, accuracy, and latency: mobile-versatile-1, mobile-low-latency-1, mobile-high-accuracy-1, and mobile-core-ml-low-latency-12. Each model has its own trade-offs and use cases, depending on the device specifications and the application requirements.

For the use case of building an ML model to reduce the amount of time spent by quality control inspectors checking for product defects, the best model to use is the AutoML Vision Edge mobile-low-latency-1 model. This model is optimized for fast inference on mobile devices, with a latency of less than 50 milliseconds on a Pixel 1 phone2. Faster defect detection is a priority for the toy manufacturer, and the factory does not have reliable Wi-Fi, so a low-latency model that can run on the device without internet connection is ideal. The mobile-low-latency-1 model also has a small size of less than 4 MB, which makes it easy to deploy and update2. The mobile-low-latency-1 model has a slightly lower accuracy than the mobile-high-accuracy-1 model, but it is still suitable for most image classification tasks2. Therefore, the AutoML Vision Edge mobile-low-latency-1 model is the best option for this use case.

References:

• AutoML Vision Edge documentation
• AutoML Vision Edge model types

Vertex AI Pipelines and AI Platform Prediction are the platform components that best suit the requirements of the data science team. Vertex AI Pipelines is a service that allows you to orchestrate and automate your machine learning workflows using pipelines. Pipelines are portable and scalable ML workflows that are based on containers. You can use Vertex AI Pipelines to schedule model retraining, use custom containers, and integrate with other Google Cloud services. AI Platform Prediction is a service that allows you to host your trained models and serve online predictions. You can use AI Platform Prediction to deploy models trained on Vertex AI or elsewhere, and benefit from features such as autoscaling, monitoring, logging, and explainability. References:

• Vertex AI Pipelines
• AI Platform Prediction

• Cost-effectiveness: User-managed notebooks in Vertex AI Workbench allow you to leverage pre-configured virtual machines with reasonable resource allocation, keeping costs lower compared to options involving managed notebooks or Dataproc clusters.
• Development flexibility: User-managed notebooks offer full control over the environment, allowing you to install additional libraries or dependencies needed for your specific EDA, preprocessing, and model training tasks. This flexibility is crucial while experimenting with different algorithms.
• BigQuery integration: The %%bigquery magic commands provide seamless integration with BigQuery within the Jupyter Notebook environment. This enables efficient querying and exploration of customer transaction data stored in BigQuery directly from the notebook, streamlining the workflow.

Other options and why they are not the best fit:

• B. Managed notebook: While managed notebooks offer an easier setup, they might have limited customization options, potentially hindering your ability to install specific libraries or tools.
• C. Dataproc Hub: Dataproc Hub focuses on running large-scale distributed workloads, and it might be overkill for your scenario involving exploratory analysis and experimentation with different algorithms. Additionally, it could incur higher costs compared to a user-managed notebook.
• D. Dataproc cluster with spark-bigquery-connector: Similar to option C, using a Dataproc cluster with the spark-bigquery-connector would be more complex and potentially more expensive than using %%bigquery magic commands within a user-managed notebook for accessing BigQuery data.

References:

• https://cloud.google.com/vertex-ai/docs/workbench/instances/bigquery
• https://cloud.google.com/vertex-ai-notebooks

Data leakage is a problem where information from outside the training dataset is used to create the model, resulting in an overly optimistic or invalid estimate of the model performance. Data leakage can occur in time series data when the temporal order of the data is not preserved during data preparation or model evaluation. For example, if the data is shuffled before splitting into train and test sets, or if future data is used to impute missing values in past data, then data leakage can occur.

One way to address data leakage in time series data is to apply nested cross-validation during model training. Nested cross-validation is a technique that allows you to perform both model selection and model evaluation in a robust way, while preserving the temporal order of the data. Nested cross-validation involves two levels of cross-validation: an inner loop for model selection and an outer loop for model evaluation. The inner loop splits the training data into k folds, trains and tunes the model on k-1 folds, and validates the model on the remaining fold. The inner loop repeats this process for each fold and selects the best model based on the validation performance. The outer loop splits the data into n folds, trains the best model from the inner loop on n-1 folds, and tests the model on the remaining fold. The outer loop repeats this process for each fold and evaluates the model performance based on the test results.

Nested cross-validation can help to avoid data leakage in time series data by ensuring that the model is trained and tested on non-overlapping data, and that the data used for validation is never seen by the model during training. Nested cross-validation can also provide a more reliable estimate of the model performance than a single train-test split or a simple cross-validation, as it reduces the variance and bias of the estimate.

References:

• Data Leakage in Machine Learning
• How to Avoid Data Leakage When Performing Data Preparation
• Classification on a single time series - prevent leakage between train and test

This option is the best way to architect the workflow, as it allows you to use event-driven and serverless components to automate the ML training process. Cloud Storage triggers are a feature that allows you to send notifications to a Pub/Sub topic when an object is created, deleted, or updated in a storage bucket. Pub/Sub is a service that allows you to publish and subscribe to messages on various topics. Pub/Sub-triggered Cloud Functions are a type of Cloud Functions that are invoked when a message is published to a specific Pub/Sub topic. Cloud Functions are a serverless platform that allows you to run code in response to events. By using these components, you can create a workflow that starts the training job on a GKE cluster as soon as a new file is available in the Cloud Storage bucket, without having to manage any servers or poll for changes. The other options are not as efficient or scalable as this option. Dataflow is a service that allows you to create and run data processing pipelines, but it is not designed to trigger ML training jobs on GKE. App Engine is a service that allows you to build and deploy web applications, but it is not suitable for polling Cloud Storage for new files, as it may incur unnecessary costs and latency. Cloud Scheduler is a service that allows you to schedule jobs at regular intervals, but it is not ideal for triggering ML training jobs based on data availability, as it may miss some files or run unnecessary jobs. References:

• Cloud Storage triggers documentation
• Pub/Sub documentation
• Pub/Sub-triggered Cloud Functions documentation
• Cloud Functions documentation
• Kubeflow Pipelines documentation

• Quantization is a technique that reduces the numerical precision of the weights and activations of a neural network, which can improve the inference speed and reduce the memory footprint of the model1.
• Reducing the floating point precision from tf.float64 to tf.float16 can potentially halve the latency and memory usage of the model, while having minimal impact on the accuracy2.
• Increasing the dropout rate to 0.8 in either mode would not affect the latency, but would likely degrade the performance of the model significantly, as dropout is a regularization technique that randomly drops out units during training to prevent overfitting3.
• Switching from CPU to GPU serving may or may not improve the latency, depending on the hardware specifications and the model complexity, but it would also incur additional costs and complexity for deployment4

 In this scenario, the goal is to speed up model training for a CNN-based architecture on Google Cloud. The code does not include any manual device placement and has not been wrapped in Estimator model-level abstraction. Given these constraints, the best environment to train the model on would be a Deep Learning VM with an n1-standard-2 machine and 1 GPU with all libraries pre-installed. Option C is the correct answer.

Option C: A Deep Learning VM with an n1-standard-2 machine and 1 GPU with all libraries pre-installed. This option is the most suitable for the scenario because it provides a ready-to-use environment for deep learning on Google Cloud. A Deep Learning VM is a specialized VM image that is pre-installed with popular deep learning frameworks such as TensorFlow, PyTorch, Keras, and more. A Deep Learning VM also comes with NVIDIA GPU drivers and CUDA libraries that enable GPU acceleration for model training. A Deep Learning VM can be easily configured and launched from the Google Cloud Console or the Cloud SDK. An n1-standard-2 machine is a general-purpose machine type that provides 2 vCPUs and 7.5 GB of memory. This machine type can be sufficient for running a CNN-based architecture. A GPU is a specialized hardware accelerator that can speed up the computation of matrix operations and convolutions, which are common in CNN-based architectures. By using a Deep Learning VM with an n1-standard-2 machine and 1 GPU, the model training can be significantly faster than on an on-premises CPU-only infrastructure.

Option A: A VM on Compute Engine and 1 TPU with all dependencies installed manually. This option is not suitable for the scenario because it requires manual installation of dependencies and device placement. A TPU is a custom-designed ASIC that can provide high performance and efficiency for TensorFlow models. However, to use a TPU, the code needs to include manual device placement and be wrapped in Estimator model-level abstraction. Moreover, to use a TPU, the dependencies such as TensorFlow, Cloud TPU Client, and Cloud Storage need to be installed manually on the VM. This option can be complex and time-consuming to set up and may not be compatible with the existing code.

Option B: A VM on Compute Engine and 8 GPUs with all dependencies installed manually. This option is not suitable for the scenario because it requires manual installation of dependencies and may not be cost-effective. While using 8 GPUs can provide high parallelism and speed for model training, it also increases the cost and complexity of the environment. Moreover, to use GPUs, the dependencies such as NVIDIA GPU drivers, CUDA libraries, and deep learning frameworks need to be installed manually on the VM. This option can be tedious and error-prone to set up and may not be necessary for the scenario.

Option D: A Deep Learning VM with more powerful CPU e2-highcpu-16 machines with all libraries pre-installed. This option is not suitable for the scenario because it does not leverage GPU acceleration for model training. While using more powerful CPU machines can provide more compute resources and memory for model training, it may not be as fast and efficient as using GPU machines. CPU machines are not optimized for matrix operations and convolutions, which are common in CNN-based architectures. Moreover, using more powerful CPU machines can also increase the cost of the environment. This option can be suboptimal and wasteful for the scenario.

References:

• Deep Learning VM Image documentation
• Compute Engine documentation
• Cloud TPU documentation
• Machine types documentation
• GPUs on Compute Engine documentation

The best business metric to track to measure the model’s performance is user engagement as measured by the number of battles played daily per user. This metric reflects the main goal of the model, which is to enhance the user experience and satisfaction by creating balanced and fair battles. If the model is successful, it should increase the user retention and loyalty, as well as the word-of-mouth and referrals. This metric is also easy to measure and interpret, as it can be directly obtained from the user activity data.

The other options are not optimal for the following reasons:

• A. Average time players wait before being assigned to a team is not a good metric, as it does not capture the quality or outcome of the battles. It only measures the efficiency of the model, which is not the primary objective. Moreover, this metric can be influenced by external factors, such as the availability and demand of players, the network latency, and the server capacity.
• B. Precision and recall of assigning players to teams based on their predicted versus actual ability is not a good metric, as it is difficult to measure and interpret. It requires having a reliable and consistent way of estimating the player’s ability, which can be subjective and dynamic. It also requires having a ground truth label for each assignment, which can be costly and impractical to obtain. Moreover, this metric does not reflect the user feedback or satisfaction, which is the ultimate goal of the model.
• D. Rate of return as measured by additional revenue generated minus the cost of developing a new model is not a good metric, as it is not directly related to the model’s performance. It measures the profitability of the model, which is a secondary objective. Moreover, this metric can be affected by many other factors, such as the market conditions, the pricing strategy, the marketing campaigns, and the competition.

References:

• Professional ML Engineer Exam Guide
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate
• Google Cloud launches machine learning engineer certification
• How to measure user engagement
• How to choose the right metrics for your machine learning model

 Before building an insurance approval model, an ML engineer should consider the factors of traceability, reproducibility, and explainability, as these are important aspects of responsible AI and fairness in a regulated domain. Traceability is the ability to track the provenance and lineage of the data, models, and decisions throughout the ML lifecycle. It helps to ensure the quality, reliability, and accountability of the ML system, and to comply with the regulatory and ethical standards. Reproducibility is the ability to recreate the same results and outcomes using the same data, models, and parameters. It helps to verify the validity, consistency, and robustness of the ML system, and to debug and improve the performance. Explainability is the ability to understand and interpret the logic, behavior, and outcomes of the ML system. It helps to increase the transparency, trust, and confidence of the ML system, and to identify and mitigate any potential biases, errors, or risks. The other options are not as relevant or comprehensive as this option. Redaction is the process of removing sensitive or confidential information from the data or documents, but it is not a factor that the ML engineer should consider before building the model, as it is more related to the data preparation and protection. Federated learning is a technique that allows training ML models on decentralized data without transferring the data to a central server, but it is not a factor that the ML engineer should consider before building the model, as it is more related to the model architecture and privacy preservation. Differential privacy is a method that adds noise to the data or the model outputs to protect the individual privacy of the data subjects, but it is not a factor that the ML engineer should consider before building the model, as it is more related to the model evaluation and deployment. References:

• Responsible AI documentation
• Traceability documentation
• Reproducibility documentation
• Explainability documentation

• Explanation: TensorFlow Extended (TFX) is a platform for building end-to-end machine learning pipelines using TensorFlow. TFX provides a set of components that can be orchestrated using either the TFX SDK or Kubeflow Pipelines. TFX components can handle different aspects of the pipeline, such as data ingestion, data validation, data transformation, model training, model evaluation, model serving, and more. TFX components can also leverage other Google Cloud services, such as BigQuery, Dataflow, and Vertex AI.
• Why not A: Using the Kubeflow Pipelines SDK to implement the pipeline is a valid option, but using the BigQueryJobOp component to run the preprocessing script is not optimal. This would require writing and maintaining a separate SQL script for data transformation, which could introduce inconsistencies and errors. It would also make it harder to reuse the same preprocessing logic for both training and serving.
• Why not B: Using the Kubeflow Pipelines SDK to implement the pipeline is a valid option, but using the DataflowPythonJobOp component to preprocess the data is not optimal. This would require writing and maintaining a separate Python script for data transformation, which could introduce inconsistencies and errors. It would also make it harder to reuse the same preprocessing logic for both training and serving.
• Why not D: Using the TensorFlow Extended SDK to implement the pipeline is a valid option, but implementing the preprocessing steps as part of the input_fn of the model is not optimal. This would make the preprocessing logic tightly coupled with the model code, which could reduce modularity and flexibility. It would also make it harder to reuse the same preprocessing logic for both training and serving.

• Option A is correct because using Vertex AI Workbench user-managed notebooks to generate the report is the best way to quickly determine whether the data is suitable for model development, and to create a one-time report that includes both informative visualizations of data distributions and more sophisticated statistical analyses to share with other ML engineers on your team. Vertex AI Workbench is a service that allows you to create and use notebooks for ML development and experimentation. You can use Vertex AI Workbench to connect to your BigQuery table, query and analyze the data using SQL or Python, and create interactive charts and plots using libraries such as pandas, matplotlib, or seaborn. You can also use Vertex AI Workbench to perform more advanced data analysis, such as outlier detection, feature engineering, or hypothesis testing, using libraries such as TensorFlow Data Validation, TensorFlow Transform, or SciPy. You can export your notebook as a PDF or HTML file, and share it with your team. Vertex AI Workbench provides maximum flexibility to create your report, as you can use any code or library that you want, and customize the report as you wish.
• Option B is incorrect because using Google Data Studio to create the report is not the most flexible way to quickly determine whether the data is suitable for model development, and to create a one-time report that includes both informative visualizations of data distributions and more sophisticated statistical analyses to share with other ML engineers on your team. Google Data Studio is a service that allows you to create and share interactive dashboards and reports using data from various sources, such as BigQuery, Google Sheets, or Google Analytics. You can use Google Data Studio to connect to your BigQuery table, explore and visualize the data using charts, tables, or maps, and apply filters, calculations, or aggregations to the data. However, Google Data Studio does not support more sophisticated statistical analyses, such as outlier detection, feature engineering, or hypothesis testing, which may be useful for model development. Moreover, Google Data Studio is more suitable for creating recurring reports that need to be updated frequently, rather than one-time reports that are static.
• Option C is incorrect because using the output from TensorFlow Data Validation on Dataflow to generate the report is not the most efficient way to quickly determine whether the data is suitable for model development, and to create a one-time report that includes both informative visualizations of data distributions and more sophisticated statistical analyses to share with other ML engineers on your team. TensorFlow Data Validation is a library that allows you to explore, validate, and monitor the quality of your data for ML. You can use TensorFlow Data Validation to compute descriptive statistics, detect anomalies, infer schemas, and generate data visualizations for your data. Dataflow is a service that allows you to create and run scalable data processing pipelines using Apache Beam. You can use Dataflow to run TensorFlow Data Validation on large datasets, such as those stored in BigQuery. However, this option is not very efficient, as it involves moving the data from BigQuery to Dataflow, creating and running the pipeline, and exporting the results. Moreover, this option does not provide maximum flexibility to create your report, as you are limited by the functionalities of TensorFlow Data Validation, and you may not be able to customize the report as you wish.
• Option D is incorrect because using Dataprep to create the report is not the most flexible way to quickly determine whether the data is suitable for model development, and to create a one-time report that includes both informative visualizations of data distributions and more sophisticated statistical analyses to share with other ML engineers on your team. Dataprep is a service that allows you to explore, clean, and transform your data for analysis or ML. You can use Dataprep to connect to your BigQuery table, inspect and profile the data using histograms, charts, or summary statistics, and apply transformations, such as filtering, joining, splitting, or aggregating, to the data. However, Dataprep does not support more sophisticated statistical analyses, such as outlier detection, feature engineering, or hypothesis testing, which may be useful for model development. Moreover, Dataprep is more suitable for creating data preparation workflows that need to be executed repeatedly, rather than one-time reports that are static.

References:

• Vertex AI Workbench documentation
• Google Data Studio documentation
• TensorFlow Data Validation documentation
• Dataflow documentation
• Dataprep documentation
• [BigQuery documentation]
• [pandas documentation]
• [matplotlib documentation]
• [seaborn documentation]
• [TensorFlow Transform documentation]
• [SciPy documentation]
• [Apache Beam documentation]

The best option for implementing the processing steps so that they run at serving time, minimizing code changes and infrastructure maintenance, and deploying the model into production as quickly as possible, is to use the Predictor interface to implement a custom prediction routine. Build the custom container, upload the container to Vertex AI Model Registry, and deploy it to a Vertex AI endpoint. This option allows you to leverage the power and simplicity of Vertex AI to serve your XGBoost model with minimal effort and customization. Vertex AI is a unified platform for building and deploying machine learning solutions on Google Cloud. Vertex AI can deploy a trained XGBoost model to an online prediction endpoint, which can provide low-latency predictions for individual instances. A custom prediction routine (CPR) is a Python script that defines the logic for preprocessing the input data, running the prediction, and postprocessing the output data. A CPR can help you customize the prediction behavior of your model, and handle complex or non-standard data formats. A CPR can also help you minimize the code changes, as you only need to write a few functions to implement the prediction logic. A Predictor interface is a class that inherits from the base class aiplatform.Predictor, and implements the abstract methods predict() and preprocess(). A Predictor interface can help you create a CPR by defining the preprocessing and prediction logic for your model. A container image is a package that contains the model, the CPR, and the dependencies. A container image can help you standardize and simplify the deployment process, as you only need to upload the container image to Vertex AI Model Registry, and deploy it to Vertex AI Endpoints. By using the Predictor interface to implement a CPR, building the custom container, uploading the container to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint, you can implement the processing steps so that they run at serving time, minimize code changes and infrastructure maintenance, and deploy the model into production as quickly as possible1.

The other options are not as good as option C, for the following reasons:

• Option A: Using FastAPI to implement an HTTP server, creating a Docker image that runs your HTTP server, and deploying it on your organization’s GKE cluster would require more skills and steps than using the Predictor interface to implement a CPR, building the custom container, uploading the container to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint. FastAPI is a framework for building web applications and APIs in Python. FastAPI can help you implement an HTTP server that can handle prediction requests and responses, and perform data preprocessing and postprocessing. A Docker image is a package that contains the model, the HTTP server, and the dependencies. A Docker image can help you standardize and simplify the deployment process, as you only need to build and run the Docker image. GKE is a service that can create and manage Kubernetes clusters on Google Cloud. GKE can help you deploy and scale your Docker image on Google Cloud, and provide high availability and performance. However, using FastAPI to implement an HTTP server, creating a Docker image that runs your HTTP server, and deploying it on your organization’s GKE cluster would require more skills and steps than using the Predictor interface to implement a CPR, building the custom container, uploading the container to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint. You would need to write code, create and configure the HTTP server, build and test the Docker image, create and manage the GKE cluster, and deploy and monitor the Docker image. Moreover, this option would not leverage the power and simplicity of Vertex AI, which can provide online prediction natively integrated with Google Cloud services2.
• Option B: Using FastAPI to implement an HTTP server, creating a Docker image that runs your HTTP server, uploading the image to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint would require more skills and steps than using the Predictor interface to implement a CPR, building the custom container, uploading the container to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint. FastAPI is a framework for building web applications and APIs in Python. FastAPI can help you implement an HTTP server that can handle prediction requests and responses, and perform data preprocessing and postprocessing. A Docker image is a package that contains the model, the HTTP server, and the dependencies. A Docker image can help you standardize and simplify the deployment process, as you only need to build and run the Docker image. Vertex AI Model Registry is a service that can store and manage your machine learning models on Google Cloud. Vertex AI Model Registry can help you upload and organize your Docker image, and track the model versions and metadata. Vertex AI Endpoints is a service that can provide online prediction for your machine learning models on Google Cloud. Vertex AI Endpoints can help you deploy your Docker image to an online prediction endpoint, which can provide low-latency predictions for individual instances. However, using FastAPI to implement an HTTP server, creating a Docker image that runs your HTTP server, uploading the image to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint would require more skills and steps than using the Predictor interface to implement a CPR, building the custom container, uploading the container to Vertex AI Model Registry, and deploying it to a Vertex AI endpoint. You would need to write code, create and configure the HTTP server, build and test the Docker image, upload the Docker image to Vertex AI Model Registry, and deploy the Docker image to Vertex AI Endpoints. Moreover, this option would not leverage the power and simplicity of Vertex AI, which can provide online prediction natively integrated with Google Cloud services2.
• Option D: Using the XGBoost prebuilt serving container when importing the trained model into Vertex AI, deploying the model to a Vertex AI endpoint, working with the backend engineers to implement the pre- and postprocessing steps in the Golang backend service would not allow you to implement the processing steps so that they run at serving time, and could increase the code changes and infrastructure maintenance. A XGBoost prebuilt serving container is a container image that is provided by Google Cloud, and contains the XGBoost framework and the dependencies. A XGBoost prebuilt serving container can help you deploy a XGBoost model without writing any code, but it also limits your customization options. A XGBoost prebuilt serving container can only handle standard data formats, such as JSON or CSV, and cannot perform any preprocessing or postprocessing on the input or output data. If your input data requires any transformation or normalization before running the prediction, you cannot use a XGBoost prebuilt serving container. A Golang backend service is a service that is implemented in Golang, a programming language that can be used for web development and system programming. A Golang backend service can help you handle the prediction requests and responses from the frontend, and communicate with the Vertex AI endpoint. However, using the XGBoost prebuilt serving container when importing the trained model into Vertex AI, deploying the model to a Vertex AI endpoint, working with the backend engineers to implement the pre- and postprocessing steps in the Golang backend service would not allow you to implement the processing steps so that they run at serving time, and could increase the code changes and infrastructure maintenance. You would need to write code, import the trained model into Vertex AI, deploy the model to a Vertex AI endpoint, implement the pre- and postprocessing steps in the Golang backend service, and test and monitor the Golang backend service. Moreover, this option would not leverage the power and simplicity of Vertex AI, which can provide online prediction natively integrated with Google Cloud services2.

References:

• Preparing for Google Cloud Certification: Machine Learning Engineer, Course 3: Production ML Systems, Week 2: Serving ML Predictions
• Google Cloud Professional Machine Learning Engineer Exam Guide, Section 3: Scaling ML models in production, 3.1 Deploying ML models to production
• Official Google Cloud Certified Professional Machine Learning Engineer Study Guide, Chapter 6: Production ML Systems, Section 6.2: Serving ML Predictions
• Custom prediction routines
• Using pre-built containers for prediction
• Using custom containers for prediction

 BigQuery is a service that allows you to store and query large amounts of data in a scalable and cost-effective way. You can use BigQuery to prepare the data for your customer churn prediction model, such as filtering, aggregating, and transforming the data. You can then associate the data with a Vertex AI dataset, which is a service that allows you to store and manage your ML data on Google Cloud. By using a Vertex AI dataset, you can easily access the data from other Vertex AI services, such as AutoML. AutoML is a service that allows you to create and train ML models without writing code. You can use AutoML to create an AutoMLTabularTrainingJob, which is a type of job that trains a classification model for tabular data, such as customer churn. By using an AutoMLTabularTrainingJob, you can benefit from the automated feature engineering, model selection, and hyperparameter tuning that AutoML provides. You can also use Vertex Explainable AI to understand how your model is making predictions, such as which features are most important and how they affect the prediction outcome. By using BigQuery, Vertex AI dataset, and AutoMLTabularTrainingJob, you can build your first model as quickly as possible while minimizing cost and complexity. References:

• BigQuery documentation
• Vertex AI dataset documentation
• AutoMLTabularTrainingJob documentation
• Vertex Explainable AI documentation
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

 Kubeflow is an open source platform for developing, orchestrating, deploying, and running scalable and portable machine learning workflows on Kubernetes. Kubeflow Pipelines is a component of Kubeflow that allows you to build and manage end-to-end machine learning pipelines using a graphical user interface or a Python-based domain-specific language (DSL). Kubeflow Pipelines can help you automate and orchestrate your machine learning workflows, and integrate with various Google Cloud services and tools1

One of the Google Cloud services that you can use with Kubeflow Pipelines is BigQuery, which is a serverless, scalable, and cost-effective data warehouse that allows you to run fast and complex queries on large-scale data. BigQuery can help you analyze and prepare your data for machine learning, and store and manage your machine learning models2

To execute a query against BigQuery as the first step in your Kubeflow pipeline, and use the results of that query as the input to the next step in your pipeline, the easiest way to do that is to use the BigQuery Query Component, which is a pre-built component that you can find in the Kubeflow Pipelines repository on GitHub. The BigQuery Query Component allows you to run a SQL query on BigQuery, and output the results as a table or a file. You can use the component’s URL to load the component into your pipeline, and specify the query and the output parameters. You can then use the output of the component as the input to the next step in your pipeline, such as a data processing or a model training step3

The other options are not as easy or feasible. Using the BigQuery console to execute your query and then save the query results into a new BigQuery table is not a good idea, as it does not integrate with your Kubeflow pipeline, and requires manual intervention and duplication of data. Writing a Python script that uses the BigQuery API to execute queries against BigQuery is not ideal, as it requires writing custom code and handling authentication and error handling. Using the Kubeflow Pipelines DSL to create a custom component that uses the Python BigQuery client library to execute queries is not optimal, as it requires creating and packaging a Docker container image for the component, and testing and debugging the component.

References: 1: Kubeflow Pipelines overview 2: BigQuery overview 3: BigQuery Query Component

• Option A is incorrect because it does not separate the training dataset into two tables based on the features, which is necessary to train the two models separately and accurately.
• Option B is incorrect because it does not separate the training dataset into two tables based on the features, and because it uses the same monitoring-config-from parameter for both models, which would not account for the different feature selections.
• Option C is incorrect because it deploys the models to two separate endpoints, which would increase the management effort and complexity of the pipeline.
• Option D is correct because it separates the training dataset into two tables based on the features, which would enable the two models to be trained separately and accurately. It also deploys both models to the same endpoint, which would simplify the pipeline and reduce the management effort. It also submits a Vertex Al Model Monitoring job with a monitoring-config-from parameter that accounts for the model IDs and training datasets, which would enable the skew detection to work properly for each model.

The most likely problem is that the tables that you created to hold your training and validation records share some records, and you may not be using all the data in your initial table. This is because the RAND() function generates a random number between 0 and 1 for each row, and the probability of a row being in both the training and validation tables is 0.2 * 0.8 = 0.16, which is not negligible. This means that some of the records that you use to validate your model are also used to train your model, which can lead to overfitting and poor generalization. Moreover, the probability of a row being in neither the training nor the validation table is 0.2 * 0.2 = 0.04, which means that you are wasting some of the data in your initial table and reducing the size of your datasets. A better way to split your data into training and validation sets is to use a hash function on a unique identifier column, such as the following queries:

CREATE OR REPLACE TABLE ‘myproject.mydataset.training‘ AS (SELECT * FROM ‘myproject.mydataset.mytable‘ WHERE MOD(FARM_FINGERPRINT(id), 10) < 8); CREATE OR REPLACE TABLE ‘myproject.mydataset.validation‘ AS (SELECT * FROM ‘myproject.mydataset.mytable‘ WHERE MOD(FARM_FINGERPRINT(id), 10) >= 8);

This way, you can ensure that each row has a fixed 80% chance of being in the training table and a 20% chance of being in the validation table, without any overlap or omission.

References:

• Professional ML Engineer Exam Guide
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate
• Google Cloud launches machine learning engineer certification
• BigQuery ML: Splitting data for training and testing
• BigQuery: FARM_FINGERPRINT function

• Option A is incorrect because using AI Platform notebooks to perform a Lasso regression analysis on your model, which will eliminate features that do not provide a strong signal, is not a suitable way to determine which customer attribute has the most predictive power for each prediction served by the model. Lasso regression is a method of feature selection that applies a penalty to the coefficients of the linear model, and shrinks them to zero for irrelevant features1. However, this method assumes that the model is linear and additive, which may not be the case for a TensorFlow model. Moreover, this method does not provide feature attributions for each prediction, but rather for the entire dataset.
• Option B is incorrect because streaming prediction results to BigQuery, and using BigQuery’s CORR(X1, X2) function to calculate the Pearson correlation coefficient between each feature and the target variable, is not a valid way to determine which customer attribute has the most predictive power for each prediction served by the model. The Pearson correlation coefficient is a measure of the linear relationship between two variables, ranging from -1 to 12. However, this method does not account for the interactions between features or the non-linearity of the model. Moreover, this method does not provide feature attributions for each prediction, but rather for the entire dataset.
• Option C is correct because using the AI Explanations feature on AI Platform, and submitting each prediction request with the ‘explain’ keyword to retrieve feature attributions using the sampled Shapley method, is the best way to determine which customer attribute has the most predictive power for each prediction served by the model. AI Explanations is a service that allows you to get feature attributions for your deployed models on AI Platform3. Feature attributions are values that indicate how much each feature contributed to the prediction for a given instance4. The sampled Shapley method is a technique that uses the Shapley value, a game-theoretic concept, to measure the contribution of each feature to the prediction5. By using AI Explanations, you can get feature attributions for each prediction request, and identify the most important features for each customer.
• Option D is incorrect because using the What-If tool in Google Cloud to determine how your model will perform when individual features are excluded, and ranking the feature importance in order of those that caused the most significant performance drop when removed from the model, is not a practical way to determine which customer attribute has the most predictive power for each prediction served by the model. The What-If tool is a tool that allows you to visualize and analyze your ML models and datasets. However, this method requires manually editing or removing features for each instance, and observing the change in the prediction. This method is not scalable or efficient, and may not capture the interactions between features or the non-linearity of the model.

References:

• Lasso regression
• Pearson correlation coefficient
• AI Explanations overview
• Feature attributions
• Sampled Shapley method
• [What-If tool overview]

Federated learning is a machine learning technique that enables organizations to train AI models on decentralized data without centralizing or sharing it1. It allows data privacy, continual learning, and better performance on end-user devices2. Federated learning works by sending the model parameters to the devices, where they are updated locally on the device’s data, and then aggregating the updated parameters on a central server to form a global model3. This way, the data never leaves the device and the model can learn from a large and diverse dataset.

Federated learning is suitable for the use case of building an ML-based biometric authentication for the bank’s mobile app that verifies a customer’s identity based on their fingerprint. Fingerprints are considered highly sensitive personal information and cannot be downloaded and stored into the bank databases. By using federated learning, the bank can train and deploy an ML model that can recognize fingerprints without compromising the data privacy of the customers. The model can also adapt to the variations and changes in the fingerprints over time and improve its accuracy and reliability. Therefore, federated learning is the best learning strategy for this use case.

Precision and recall are two common metrics for evaluating the performance of a classification model. Precision measures the proportion of positive predictions that are correct, while recall measures the proportion of positive examples that are correctly predicted. Precision and recall are inversely related, meaning that increasing one will decrease the other, and vice versa. The trade-off between precision and recall depends on the goal and the cost of the classification problem1.

For the use case of detecting whether posted images contain cars, precision is more important than recall, as the social media company wants to minimize the number of false positives, or images that are incorrectly labeled as containing cars. A high precision means that the model is confident and accurate in its positive predictions, while a low recall means that the model may miss some positive examples, or images that actually contain cars. The cost of missing some positive examples is lower than the cost of making wrong positive predictions, as the latter may affect the user experience and the reputation of the social media company.

The softmax function is a function that transforms a vector of real numbers into a probability distribution over the possible classes. The softmax function is often used as the final layer of a neural network for multi-class classification problems, as it assigns a probability to each class, and the class with the highest probability is chosen as the prediction. The softmax function is defined as:

softmax (x_i) = exp (x_i) / sum_j exp (x_j)

where x_i is the input value for class i, and softmax (x_i) is the output probability for class i.

The softmax threshold is a parameter that determines the minimum probability that a class must have to be chosen as the prediction. For example, if the softmax threshold is 0.5, then the class with the highest probability must have at least 0.5 to be selected, otherwise the prediction is none. The softmax threshold can be used to adjust the trade-off between precision and recall, as a higher threshold will increase the precision and decrease the recall, while a lower threshold will decrease the precision and increase the recall2.

For the use case of detecting whether posted images contain cars, the best way to adjust the model’s final layer softmax threshold to increase precision is to decrease the recall. This means that the softmax threshold should be increased, so that the model will only make positive predictions when it is highly confident, and avoid making false positives. By increasing the softmax threshold, the model will become more selective and accurate in its positive predictions, and improve the precision metric. Therefore, decreasing the recall is the best option for this use case.

References:

• Precision and recall - Wikipedia
• How to add a threshold in softmax scores - Stack Overflow

BigQuery is a serverless data warehouse that allows you to perform SQL queries on large-scale data. BigQuery ML is a feature of BigQuery that enables you to create and execute machine learning models using standard SQL queries. You can use BigQuery ML to train a matrix factorization model, which is a common technique for recommender systems. Matrix factorization models learn the latent factors that represent the preferences of users and the characteristics of items, and use them to predict the ratings or interactions between users and items. You can use the CREATE MODEL statement to create a matrix factorization model in BigQuery ML, and specify the matrix_factorization option as the model type. You can also use the ML.RECOMMEND function to generate recommendations for new games based on the trained model. This solution requires the least amount of coding, as you only need to write SQL queries to train and use the model. References: The answer can be verified from official Google Cloud documentation and resources related to BigQuery and BigQuery ML.

• BigQuery ML | Google Cloud
• Using matrix factorization | BigQuery ML
• ML.RECOMMEND function | BigQuery ML

Z-score normalization is a technique that transforms the values of a numeric variable into standardized units, such that the mean is zero and the standard deviation is one. Z-score normalization can help to compare variables with different scales and ranges, and to reduce the effect of outliers and skewness. The formula for z-score normalization is:

z = (x - mu) / sigma

where x is the original value, mu is the mean of the variable, and sigma is the standard deviation of the variable.

Dataflow is a service that allows you to create and run data processing pipelines on Google Cloud. You can use Dataflow to preprocess raw data prior to model training and prediction, such as applying z-score normalization on data stored in BigQuery. However, using Dataflow for this task may not be the most efficient option, as it involves reading and writing data from and to BigQuery, which can be time-consuming and costly. Moreover, using Dataflow requires manual intervention to update the pipeline whenever new training data is added.

A more efficient way to perform z-score normalization on data stored in BigQuery is to translate the normalization algorithm into SQL and use it with BigQuery. BigQuery is a service that allows you to analyze large-scale and complex data using SQL queries. You can use BigQuery to perform z-score normalization on your data using SQL functions such as AVG(), STDDEV_POP(), and OVER(). For example, the following SQL query can normalize the values of a column called temperature in a table called weather:

SELECT (temperature - AVG(temperature) OVER ()) / STDDEV_POP(temperature) OVER () AS normalized_temperature FROM weather;

By using SQL to perform z-score normalization on BigQuery, you can make the process more efficient by minimizing computation time and manual intervention. You can also leverage the scalability and performance of BigQuery to handle large and complex datasets. Therefore, translating the normalization algorithm into SQL for use with BigQuery is the best option for this use case.

AI Platform Notebooks is a service that provides managed Jupyter notebooks for data science and machine learning. You can use AI Platform Notebooks to create, run, and share your code and analysis in a collaborative and interactive environment1. BigQuery is a service that allows you to analyze large-scale and complex data using SQL queries. You can use BigQuery to stream, store, and query your data in a fast and cost-effective way2. Pandas is a popular Python library that provides data structures and tools for data analysis and manipulation. You can use pandas to create, manipulate, and visualize dataframes, which are tabular data structures with rows and columns3.

AI Platform Notebooks provides a cell magic, %%bigquery, that allows you to run SQL queries on BigQuery data and ingest the results as a pandas dataframe. A cell magic is a special command that applies to the whole cell in a Jupyter notebook. The %%bigquery cell magic can take various arguments, such as the name of the destination dataframe, the name of the destination table in BigQuery, the project ID, and the query parameters4. By using the %%bigquery cell magic, you can query the data in BigQuery with minimal code and manipulate the results with pandas in AI Platform Notebooks. This is the most convenient and efficient way to achieve your goal.

The other options are not as good as option A, because they involve more steps, more code, and more manual effort. Option B requires you to export your table as a CSV file from BigQuery to Google Drive, and then use the Google Drive API to ingest the file into your notebook instance. This option is cumbersome and time-consuming, as it involves moving the data across different services and formats. Option C requires you to download your table from BigQuery as a local CSV file, and then upload it to your AI Platform notebook instance. This option is also inefficient and impractical, as it involves downloading and uploading large files, which can take a long time and consume a lot of bandwidth. Option D requires you to use a bash cell in your AI Platform notebook to export the table as a CSV file to Cloud Storage, and then copy the data into the notebook. This option is also complex and unnecessary, as it involves using different commands and tools to move the data around. Therefore, option A is the best option for this use case.

References:

• AI Platform Notebooks documentation
• BigQuery documentation
• pandas documentation
• Using Jupyter magics to query BigQuery data

A feature cross is a synthetic feature that is obtained by combining two or more existing features, usually by taking their product or concatenation. A feature cross can help to capture the nonlinear and interaction effects between the original features, and improve the predictive performance of the model. A feature cross can be applied to different types of features, such as numeric, categorical, or geospatial features1.

For the use case of building an ML model to predict car sales in different cities around the world, the best option is to use one feature obtained as an element-wise product between binned latitude, binned longitude, and one-hot encoded car type. This option involves creating a feature cross that combines three individual features: binned latitude, binned longitude, and one-hot encoded car type. Binning is a technique that transforms a continuous numeric feature into a discrete categorical feature by dividing its range into equal intervals, or bins. One-hot encoding is a technique that transforms a categorical feature into a binary vector, where each element corresponds to a possible category, and has a value of 1 if the feature belongs to that category, and 0 otherwise. By applying binning and one-hot encoding to the latitude, longitude, and car type features, the feature cross can capture the city-specific relationships between car type and number of sales, as each combination of bins and car types can represent a different city and its preference for a certain car type. For example, the feature cross can learn that a city with a latitude bin of [40, 50], a longitude bin of [-80, -70], and a car type of SUV has a higher number of sales than a city with a latitude bin of [-10, 0], a longitude bin of [10, 20], and a car type of sedan. Therefore, using one feature obtained as an element-wise product between binned latitude, binned longitude, and one-hot encoded car type is the best option for this use case.

References:

• Feature Crosses | Machine Learning Crash Course

Vertex AI Model Monitoring is a service that allows you to monitor the performance and quality of your ML models in production. You can use Vertex AI Model Monitoring to detect changes in the input data distribution, the prediction output distribution, or the model accuracy over time. Training-serving skew detection is a feature of Vertex AI Model Monitoring that compares the statistics of the data used for training the model and the data used for serving the model. If there is a significant difference between the two data distributions, it indicates that the model might be outdated or inaccurate. By enabling training-serving skew detection for your model, you can detect model performance changes in production and trigger retraining or redeployment of your model as needed. This way, you can ensure that your model is always up-to-date and accurate, without moving your data to the cloud. References:

• Vertex AI Model Monitoring documentation
• Training-serving skew detection documentation
• Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

Vertex AI Endpoint is a service that allows you to serve your ML models online and scale them automatically. You can use Vertex AI Endpoint to deploy the custom ML model that you developed for recommending recipes to the users. You can maintain the same machine type on the endpoint, which is a single machine with 8 vCPUs and no accelerators. This machine type can optimize the costs by using the queries per second (QPS) that the model can serve. You can also configure the endpoint to enable autoscaling based on vCPU usage. Autoscaling is a feature that allows the endpoint to adjust the number of compute nodes based on the traffic demand. By enabling autoscaling based on vCPU usage, you can ensure that the endpoint can scale efficiently to the increased demand during the holiday season, without overprovisioning or underprovisioning the resources. You can also set up a monitoring job and an alert for CPU usage. Monitoring is a service that allows you to collect and analyze the metrics and logs from your Google Cloud resources. You can use Monitoring to monitor the CPU usage of your endpoint, which is an indicator of the load and performance of your model. You can also set up an alert for CPU usage, which is a feature that allows you to receive notifications when the CPU usage exceeds a certain threshold. By setting up a monitoring job and an alert for CPU usage, you can keep track of the health and status of your endpoint, and detect any issues or anomalies. If you receive an alert, you can investigate the cause by using the Monitoring dashboard, which provides a graphical interface for viewing and analyzing the metrics and logs from your endpoint. You can also use the Monitoring dashboard to troubleshoot and resolve the issues, such as adjusting the autoscaling parameters, optimizing the model, or updating the machine type. By using Vertex AI Endpoint, autoscaling, and Monitoring, you can ensure that the model can scale efficiently to the increased demand during the holiday season, and handle any issues or alerts that might arise. References:

• [Vertex AI Endpoint documentation]
• [Autoscaling documentation]
• [Monitoring documentation]
• [Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate]

 The best option for building a recommendation system without any user event data is to use simple heuristics based on content metadata. This is a type of content-based filtering, which recommends items that are similar to the ones that the user has interacted with or selected, based on their attributes. For example, if a user selects a comedy movie from the US released in 2020, the system can recommend other comedy movies from the US released in 2020 or nearby years. This approach does not require any machine learning, but it can leverage the existing metadata of the videos to provide relevant recommendations. It also allows the system to start collecting user event data, such as views, likes, ratings, etc., which can be used to train a more sophisticated machine learning model in the future, such as a collaborative filtering model or a hybrid model that combines content and collaborative information. References:

• Recommendation Systems
• Content-Based Filtering
• Collaborative Filtering
• Hybrid Recommender Systems: A Systematic Literature Review