Data Preparation for Machine Learning

A data scientist is preparing data for a machine learning model and needs to handle missing values in a numeric feature. Which of the following methods is most appropriate when the data is not missing completely at random and has potential bias?

When constructing a data ingestion pipeline for a real-time streaming application on AWS, which combination of services would provide a scalable and cost-effective solution?

A machine learning engineer is tasked with transforming a categorical feature with high cardinality (many unique values) into a numerical format suitable for model training. Which encoding method is most likely to cause the 'curse of dimensionality'?

A data scientist is preparing a dataset for training a classification model and observes a significant class imbalance. Which of the following techniques is most likely to improve model performance without introducing bias?

When choosing between object, shared file, and high-performance file storage on AWS for machine learning workloads, which of the following scenarios is most suitable for using Amazon EFS?

During model development, you discover a substantial difference in proportions of labels (DPL) between your training and validation datasets. What is the most appropriate action to take?

You're building a machine learning model using Amazon SageMaker and need to ensure data privacy and compliance with regulations such as GDPR. Which of the following techniques is most appropriate for protecting sensitive customer data during the model training process?

A company needs to combine customer data from a SQL database with website activity logs stored in S3 for a machine learning model. Which approach would ensure efficient data merging during the data preparation phase?

A data scientist is working on detecting outliers in a dataset. Which of the following techniques is most appropriate when they want to identify data points that deviate significantly from the norm in multiple dimensions?

Which of the following AWS services would you primarily use to facilitate the process of labeling a large image dataset for training a computer vision model?

When preparing a dataset for a fraud detection model, which data transformation technique would be most suitable to convert categorical variables like 'city' into a numerical format that the model can process effectively?

During data preparation, an ML engineer identifies a skewed distribution in a numerical feature. Which transformation technique would you employ to make the distribution more symmetrical and improve model performance?

A machine learning engineer aims to implement feature binning on a continuous numerical feature. What is the primary reason for applying this technique?

A machine learning team is preparing data for a sentiment analysis model. They need to label a large dataset of customer reviews but want to minimize manual effort. Which approach would be most effective for this task?

A financial institution is developing a model to predict loan defaults. They are concerned about potential bias in their training data related to demographic features. Which technique could they use to assess whether pre-training bias exists in their dataset?

A healthcare provider is training a model to predict patient readmission rates. They need to ensure that the model complies with HIPAA regulations and protects patient privacy during the data preparation phase. Which strategy should they prioritize?

An e-commerce company wants to use real-time website activity data to personalize product recommendations. Which AWS service is best suited for ingesting and processing streaming data for this use case?

Consider a scenario where raw data contains a feature 'transaction_amount' with several extreme outliers. Which of the following data preparation techniques would be MOST effective in mitigating the impact of these outliers on a machine learning model's performance?

You're building a credit risk model and discover that your training dataset contains a disproportionately low number of defaults (highly imbalanced data). What data preparation technique would be LEAST suitable to address this imbalance?

A machine learning model is underperforming due to inconsistent data quality across different data sources. Specifically, some values for a 'product_price' feature are stored in USD while others are in EUR. What data preparation step is MOST crucial?

Which of the following is a critical consideration when choosing between real-time and batch deployment strategies for a machine learning model?

A data scientist needs to automate the build process for their machine learning application within a CI/CD pipeline. Which AWS service is most suitable for this task?

Which deployment strategy minimizes downtime and risk by gradually shifting user traffic from an old version of a model to a new version?

What is the primary benefit of using Infrastructure as Code (IaC) tools like AWS CloudFormation or AWS CDK in machine learning deployments?

A machine learning engineer is observing high latency in a deployed model. Which metric should they primarily monitor to diagnose the issue?

Which AWS service can be used to establish a private and isolated network environment for deploying SageMaker endpoints, enhancing security and compliance?

In the context of MLOps, what is the purpose of implementing unit tests, integration tests, and end-to-end tests for machine learning models and code?

Which of the options are key performance metrics for ML infrastructure?

A company wants to track the costs associated with different machine learning projects. Which AWS service feature helps in cost tracking and allocation using metadata?

What is the purpose of 'least privilege access' when configuring IAM roles and policies for machine learning workflows on AWS?

A machine learning model's performance degrades significantly after deployment due to changes in the incoming data. Which of the following techniques can be used detect these changes?

Which AWS service should a machine learning engineer use to monitor and log all API calls made to their AWS resources, aiding in security auditing and compliance?

When should a machine learning engineer consider using SageMaker Savings Plans for deploying ML models?

Which of the following factors should be considered when right-sizing instances for machine learning inference?

What is the primary benefit of using multi-model endpoints in SageMaker?

Which of the following is NOT a primary consideration when selecting a deployment infrastructure for a machine learning model within an existing AWS architecture?

When deciding between real-time versus batch model serving strategies, which scenario would most likely benefit from a real-time serving approach?

What is the primary advantage of using Infrastructure as Code (IaC) for managing machine learning deployments on AWS?

Which of the following is a key benefit of containerizing machine learning models for deployment?

When configuring VPC networking for SageMaker, which of the following is a primary security benefit?

Which of the following is a critical consideration when selecting metrics for auto scaling a SageMaker endpoint?

In the context of CI/CD for machine learning, what is the primary purpose of automated testing within the pipeline?

Which of the following is a key consideration when designing a retraining strategy for a deployed machine learning model?

What is the role of a version control system (like Git) in a machine learning CI/CD pipeline?

When choosing a deployment orchestrator for machine learning models on AWS, which of the following factors should be prioritized?

In the context of multi-model deployments, what is a common strategy for efficiently serving multiple models from a single endpoint?

When deploying machine learning models to edge devices, what is a primary optimization goal?

What is a typical trade-off to consider between performance and cost when deploying ML models?

How do you automate resource provisioning for ML model deployment?

You aim to update your machine learning model in production without disrupting the service. Which deployment strategy is MOST suitable?

Data Ingestion

Data Formats

AWS Data Storage Services

Streaming Data Ingestion

AWS Storage Trade-offs

Data Extraction Methods

Data Transformation

Feature Engineering

Encoding Techniques

Bias Mitigation Strategies

Non-Validated Formats

SageMaker Ground Truth

Class Imbalance (CI)

Difference in Proportions of Labels (DPL)

Synthetic Data Generation

Selection Bias

Measurement Bias

Dataset Splitting

Dataset Shuffling

Dataset Augmentation

Epoch

Batch Size

Early Stopping

Distributed Training

Regularization

Feature Selection

Ensembling

F1 Score

Confusion Matrix

Heat Maps

Accuracy

Recall

RMSE

AUC

Identify Overfitting and Underfitting

Batch Processing

CI/CD

Canary Deployment

Blue/Green Deployment

Auto Scaling

AWS CloudFormation

Amazon ECS

Amazon EKS

Amazon ECR

Amazon SageMaker

Model Drift

AWS CloudTrail

IAM Role

Least Privilege