PMLE practice questions 281–290

1. Question 281 · 364 in bank

   You need to design an architecture that serves asynchronous predictions to determine whether a particular mission-critical machine part will fail. Your system collects data from multiple sensors from the machine. You want to build a model that will predict a failure in the next N minutes, given the average of each sensor’s data from the past 12 hours. How should you design the architecture?

   • A1. HTTP requests are sent by the sensors to your ML model, which is deployed as a microservice and exposes a REST API for prediction 2. Your application queries a Vertex AI endpoint where you deployed your model. 3. Responses are received by the caller application as soon as the model produces the prediction.
   • B1. Events are sent by the sensors to Pub/Sub, consumed in real time, and processed by a Dataflow stream processing pipeline. 2. The pipeline invokes the model for prediction and sends the predictions to another Pub/Sub topic. 3. Pub/Sub messages containing predictions are then consumed by a downstream system for monitoring.
   • C1. Export your data to Cloud Storage using Dataflow. 2. Submit a Vertex AI batch prediction job that uses your trained model in Cloud Storage to perform scoring on the preprocessed data. 3. Export the batch prediction job outputs from Cloud Storage and import them into Cloud SQL.
   • D1. Export the data to Cloud Storage using the BigQuery command-line tool 2. Submit a Vertex AI batch prediction job that uses your trained model in Cloud Storage to perform scoring on the preprocessed data. 3. Export the batch prediction job outputs from Cloud Storage and import them into BigQuery.

   Show answer

   Answer: B

   Explanation

   B is most likely . if you search asynchronous on this page. it appears in the question wants to focus on online prediction with asynchronous mode. https://cloud.google.com/architecture/minimizing-predictive-serving-latency-in-machine-learning#online_real-time_prediction and the question is the same as what has been explained in this section obviously. it is as below. Predictive maintenance: asynchronously predicting whether a particular machine part will fail in the next N minutes, given the averages of the sensor's data in the past 30 minutes. afte that, you can take a closer look at figure3 and read what it try to describle C and D it is the offline solution but you opt to use different tools. https://cloud.google.com/architecture/minimizing-predictive-serving-latency-in-machine-learning#offline_batch_prediction

   Source: Adapted from community discussion — John_Pongthorn

   References

   • Google Cloud documentation

   Next question

2. Question 282 · 364 in bank

   Your company manages an application that aggregates news articles from many different online sources and sends them to users. You need to build a recommendation model that will suggest articles to readers that are similar to the articles they are currently reading. Which approach should you use?

   • ACreate a collaborative filtering system that recommends articles to a user based on the user’s past behavior.
   • BEncode all articles into vectors using word2vec, and build a model that returns articles based on vector similarity.
   • CBuild a logistic regression model for each user that predicts whether an article should be recommended to a user.
   • DManually label a few hundred articles, and then train an SVM classifier based on the manually classified articles that categorizes additional articles into their respective categories.

   Show answer

   Answer: B

   Explanation

   Option A, creating a collaborative filtering system, may not be ideal for this use case because it relies on user behavior data, which may not be available or sufficient for new users or for users who have not interacted with the system much. Option C, building a logistic regression model for each user, may not be scalable because it requires building a separate model for each user, which can become difficult to manage as the number of users increases. Option D, manually labeling articles and training an SVM classifier, may not be as effective as the word2vec approach because it relies on manual labeling, which can be time-consuming and may not capture the full semantic meaning of the articles. Additionally, SVMs may not be as effective as neural network-based approaches like word2vec for capturing complex relationships between words and articles.

   Source: Adapted from community discussion — tavva_prudhvi

   References

   • Google Cloud documentation

   Next question

3. Question 283 · 364 in bank

   You work for a large social network service provider whose users post articles and discuss news. Millions of comments are posted online each day, and more than 200 human moderators constantly review comments and flag those that are inappropriate. Your team is building an ML model to help human moderators check content on the platform. The model scores each comment and flags suspicious comments to be reviewed by a human. Which metric(s) should you use to monitor the model’s performance?

   • ANumber of messages flagged by the model per minute
   • BNumber of messages flagged by the model per minute confirmed as being inappropriate by humans.
   • CPrecision and recall estimates based on a random sample of 0.1% of raw messages each minute sent to a human for review
   • DPrecision and recall estimates based on a sample of messages flagged by the model as potentially inappropriate each minute

   Show answer

   Answer: D

   Explanation

   Precision and recall are critical metrics for evaluating the performance of classification models, especially in contexts where both the accuracy of positive predictions (precision) and the ability to identify all positive instances (recall) are important. In this case: Precision (the proportion of messages flagged by the model as inappropriate that were actually inappropriate) helps ensure that the model minimizes the burden on human moderators by not flagging too many false positives, which could overwhelm them. Recall (the proportion of actual inappropriate messages that were correctly flagged by the model) ensures that the model is effective at catching as many inappropriate messages as possible, reducing the risk of harmful content being missed.

   Source: Adapted from community discussion — ludovikush

   References

   • Google Cloud documentation
   • Google Cloud documentation

   Next question

4. Question 284 · 364 in bank

   You are a lead ML engineer at a retail company. You want to track and manage ML metadata in a centralized way so that your team can have reproducible experiments by generating artifacts. Which management solution should you recommend to your team?

   • AStore your tf.logging data in BigQuery.
   • BManage all relational entities in the Hive Metastore.
   • CStore all ML metadata in Google Cloud’s operations suite.
   • DManage your ML workflows with Vertex ML Metadata.

   Show answer

   Answer: D

   References

   • Google Cloud documentation
   • developers.google.com documentation

   Next question

5. Question 285 · 364 in bank

   You have been given a dataset with sales predictions based on your company’s marketing activities. The data is structured and stored in BigQuery, and has been carefully managed by a team of data analysts. You need to prepare a report providing insights into the predictive capabilities of the data. You were asked to run several ML models with different levels of sophistication, including simple models and multilayered neural networks. You only have a few hours to gather the results of your experiments. Which Google Cloud tools should you use to complete this task in the most efficient and self-serviced way?

   • AUse BigQuery ML to run several regression models, and analyze their performance.
   • BRead the data from BigQuery using Dataproc, and run several models using SparkML.
   • CUse Vertex AI Workbench user-managed notebooks with scikit-learn code for a variety of ML algorithms and performance metrics.
   • DTrain a custom TensorFlow model with Vertex AI, reading the data from BigQuery featuring a variety of ML algorithms.

   Show answer

   Answer: A

   Explanation

   All deep neural networks are multilayered neural networks, but not all multilayered neural networks are necessarily deep. The term "deep" is used to emphasize the depth of the network in the context of having many hidden layers, which has been shown to be effective for learning hierarchical representations of complex patterns in data. Hence BQ allows creation of DNNs (https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create-dnn-models) it should be A.

   Source: Adapted from community discussion — iieva

   References

   • Google Cloud documentation

   Next question

6. Question 286 · 364 in bank

   You are an ML engineer at a bank. You have developed a binary classification model using AutoML Tables to predict whether a customer will make loan payments on time. The output is used to approve or reject loan requests. One customer’s loan request has been rejected by your model, and the bank’s risks department is asking you to provide the reasons that contributed to the model’s decision. What should you do?

   • AUse local feature importance from the predictions.
   • BUse the correlation with target values in the data summary page.
   • CUse the feature importance percentages in the model evaluation page.
   • DVary features independently to identify the threshold per feature that changes the classification.

   Show answer

   Answer: A

   Explanation

   To access local feature importance in AutoML Tables, you can use the "Explain" feature, which shows the contribution of each feature to the prediction for a specific example. This will help you identify the most important features that contributed to the loan request being rejected. Option B, using the correlation with target values in the data summary page, may not provide the most accurate explanation as it looks at the overall correlation between the features and target variable, rather than the contribution of each feature to a specific prediction. Option C, using the feature importance percentages in the model evaluation page, may not provide a sufficient explanation for the specific prediction, as it shows the importance of each feature across all predictions, rather than for a specific prediction. Option D, varying features independently to identify the threshold per feature that changes the classification, is not recommended as it can be time-consuming and does not provide a clear explanation for why the loan request was rejected

   Source: Adapted from community discussion — shankalman717

   References

   • Google Cloud documentation
   • Google Cloud documentation

   Next question

7. Question 287 · 364 in bank

   You work for a magazine distributor and need to build a model that predicts which customers will renew their subscriptions for the upcoming year. Using your company’s historical data as your training set, you created a TensorFlow model and deployed it to AI Platform. You need to determine which customer attribute has the most predictive power for each prediction served by the model. What should you do?

   • AUse AI Platform notebooks to perform a Lasso regression analysis on your model, which will eliminate features that do not provide a strong signal.
   • BStream prediction results to BigQuery. Use BigQuery’s CORR(X1, X2) function to calculate the Pearson correlation coefficient between each feature and the target variable.
   • CUse the AI Explanations feature on AI Platform. Submit each prediction request with the ‘explain’ keyword to retrieve feature attributions using the sampled Shapley method.
   • DUse the What-If tool in Google Cloud to determine how your model will perform when individual features are excluded. Rank the feature importance in order of those that caused the most significant performance drop when removed from the model.

   Show answer

   Answer: C

   Explanation

   This is from the doc you provided: "Feature attribution is supported for all types of models (both AutoML and custom-trained), frameworks (TensorFlow, scikit, XGBoost), and modalities (images, text, tabular, video)." https://cloud.google.com/vertex-ai/docs/explainable-ai/overview#supported_model_types_2

   Source: Adapted from community discussion — mil_spyro

   References

   • Google Cloud documentation
   • Google Cloud documentation

   Next question

8. Question 288 · 364 in bank

   You are developing ML models with AI Platform for image segmentation on CT scans. You frequently update your model architectures based on the newest available research papers, and have to rerun training on the same dataset to benchmark their performance. You want to minimize computation costs and manual intervention while having version control for your code. What should you do?

   • AUse Cloud Functions to identify changes to your code in Cloud Storage and trigger a retraining job.
   • BUse the gcloud command-line tool to submit training jobs on AI Platform when you update your code.
   • CUse Cloud Build linked with Cloud Source Repositories to trigger retraining when new code is pushed to the repository.
   • DCreate an automated workflow in Cloud Composer that runs daily and looks for changes in code in Cloud Storage using a sensor.

   Show answer

   Answer: C

   Explanation

   ANS:C CI/CD for Kubeflow pipelines. At the heart of this architecture is Cloud Build, infrastructure. Cloud Build can import source from Cloud Source Repositories, GitHub, or Bitbucket, and then execute a build to your specifications, and produce artifacts such as Docker containers or Python tar files.

   Source: Adapted from community discussion — celia20200410

   References

   • Google Cloud documentation

   Next question

9. Question 289 · 364 in bank

   You are working on a binary classification ML algorithm that detects whether an image of a classified scanned document contains a company’s logo. In the dataset, 96% of examples don’t have the logo, so the dataset is very skewed. Which metrics would give you the most confidence in your model?

   • AF-score where recall is weighed more than precision
   • BRMSE
   • CF1 score
   • DF-score where precision is weighed more than recall

   Show answer

   Answer: A

   Explanation

   In this scenario, the dataset is highly imbalanced, where most of the examples do not have the company's logo. Therefore, accuracy could be misleading as the model can have high accuracy by simply predicting that all images do not have the logo. F1 score is a good metric to consider in such cases, as it takes both precision and recall into account. However, since the dataset is highly skewed, we should weigh recall more than precision to ensure that the model is correctly identifying the images that do have the logo. Therefore, F-score where recall is weighed more than precision is the best metric to evaluate the performance of the model in this scenario. Option B (RMSE) is not applicable to this classification problem, and option D (F-score where precision is weighed more than recall) is not suitable for highly skewed datasets.

   Source: Adapted from community discussion — tavva_prudhvi

   Next question

10. Question 290 · 364 in bank

    You work on the data science team for a multinational beverage company. You need to develop an ML model to predict the company’s profitability for a new line of naturally flavored bottled waters in different locations. You are provided with historical data that includes product types, product sales volumes, expenses, and profits for all regions. What should you use as the input and output for your model?

    • AUse latitude, longitude, and product type as features. Use profit as model output.
    • BUse latitude, longitude, and product type as features. Use revenue and expenses as model outputs.
    • CUse product type and the feature cross of latitude with longitude, followed by binning, as features. Use profit as model output.
    • DUse product type and the feature cross of latitude with longitude, followed by binning, as features. Use revenue and expenses as model outputs.

    Show answer

    Answer: C

    Explanation

    Option C is the best option because it takes into account both the product type and location, which can affect profitability. Binning the feature cross of latitude and longitude can help capture the nonlinear relationship between location and profitability, and using profit as the model output is appropriate because it's the target variable we want to predict.

    Source: Adapted from community discussion — tavva_prudhvi

    References

    • Google developer documentation
    • Google developer documentation
    • Google developer documentation

    Next question

Keyboard:J Kmove between questionsA–Dselect an answerRshow or hide the answer← →change page

PMLE practice, question 281. 364 questions available.