MLS-C01 Practice Test 14 — 15 Questions

Question 1

A data scientist needs to perform feature scaling for a dataset containing numerical features with different units (e.g., age in years and income in dollars). Which scaling method is most appropriate when the algorithm assumes data is normally distributed?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Standardization (Z-score normalization). Standardization (Z-score normalization) transforms features to have mean 0 and standard deviation 1, which is suitable for algorithms that assume normally distributed data. Option B (Log transformation) is used to reduce skewness, not for scaling when normality is already assumed. Option C (Min-Max scaling) scales to a fixed range (e.g., 0-1) but does not preserve normal distribution properties. Option D (Robust scaling) uses median and IQR, which is robust to outliers but does not assume normality.

Answer

Log transformation

Answer

Min-Max scaling

Answer

Robust scaling (using median and IQR)

Question 2

A data scientist is training a linear regression model on a dataset with 100 features. The model shows high variance on the test set. Which action is MOST likely to reduce overfitting?

Accepted Answer

Apply L2 regularization (Ridge regression). High variance (overfitting) means the model is too complex and fits noise in the training data. L2 regularization (Ridge regression) adds a penalty proportional to the square of the coefficients, shrinking them toward zero and reducing model complexity. This directly counteracts overfitting by preventing the model from relying too heavily on any single feature.

Answer

Use a more complex model like XGBoost

Answer

Increase the number of training iterations

Answer

Add more feature engineering to increase model complexity

Question 3

A data scientist is using SageMaker Ground Truth to create a labeled dataset for object detection. After the labeling job completes, the scientist notices that the output manifest file contains incorrect labels. What is the most efficient way to correct these labels?

Accepted Answer

Create an incremental labeling job that includes only the mislabeled items.. SageMaker Ground Truth supports incremental labeling jobs that allow you to provide a new manifest with only mislabeled items, and the job will correct only those labels without re-labeling correctly labeled data. Option B is wrong because deleting and starting over is inefficient and loses all progress. Option C is wrong because the SageMaker console does not allow direct editing of manifest files; labels are fixed only through re-labeling. Option D is wrong because it would re-label the entire dataset, wasting time and resources.

Answer

Delete the labeling job and start over with a different set of workers.

Answer

Use the SageMaker console to edit the incorrect labels directly in the manifest file.

Answer

Create a new labeling job with the same dataset and manually verify all labels.

Question 4

A machine learning team is using Amazon SageMaker to train models on a large dataset stored in Amazon S3. The dataset is 5 TB in size and is partitioned by date. The team wants to minimize data transfer costs and reduce training time by caching frequently accessed data locally on the training instances. The training instances are EC2 instances with attached Amazon EBS volumes. The team is considering using SageMaker Pipe mode to stream data directly from S3, but they are concerned about network bandwidth. Which approach should the team use to optimize data loading for training?

Accepted Answer

Use Amazon FSx for Lustre as a high-performance file system linked to the S3 bucket, and mount it on the training instances.. Amazon FSx for Lustre is purpose-built for high-performance, low-latency access to large datasets and can be linked to an S3 bucket so that data is lazily loaded and cached on the file system. Mounting it on SageMaker training instances gives the team a POSIX-compliant, high-throughput file system that caches frequently accessed data locally, minimizing repeated S3 transfers and reducing training time. This directly addresses the 5 TB dataset and the concern about network bandwidth.

Answer

Use SageMaker File mode with Amazon EFS, which allows multiple training instances to share the same file system and caches data from S3.

Answer

Increase the size of the EBS volumes attached to the training instances and copy the entire dataset to the volumes before training.

Answer

Use SageMaker Pipe mode to stream data from S3 directly to the training algorithm, which automatically caches data in memory.

Question 5

During EDA, a data scientist finds that a categorical feature 'city' has 500 unique values but only 10 cities account for 90% of the data. What is a recommended way to handle the rare categories?

Accepted Answer

Group rare categories into a single 'Other' category.. Grouping rare categories into a single 'Other' bucket is the recommended approach because it retains the signal from the 10 dominant cities while collapsing the long tail of 490 sparse categories into one manageable feature. This reduces dimensionality, prevents overfitting on categories with very few samples, and keeps the model from assigning spurious weights to noise. It is a standard feature-engineering technique for high-cardinality categorical variables.

Answer

Apply label encoding to all categories.

Answer

One-hot encode all 500 categories.

Answer

Drop all rows with rare categories.

Question 6

A company uses AWS Glue Data Catalog to manage metadata for its data lake on Amazon S3. The data lake contains terabytes of data in CSV format. The data engineering team wants to improve query performance in Amazon Athena and reduce costs. Which actions should the team take? (Select THREE.)

Accepted Answer

Compress the data using Snappy or GZIP.. Option B is correct because compressing CSV data with Snappy or GZIP reduces the number of bytes Athena must read from Amazon S3, and since Athena charges based on data scanned, this directly lowers query cost and improves performance. Option C is correct because partitioning the data by commonly filtered columns (e.g., date or region) lets Athena use partition pruning to scan only the relevant S3 prefixes instead of the entire dataset, dramatically reducing both latency and bytes scanned. Option D is correct because converting CSV to Parquet, a columnar format, allows Athena to read only the columns referenced in a query and benefits from better compression and predicate pushdown, which improves performance and reduces cost. Option A is not correct because creating views only simplifies query authoring and does not by itself reduce the amount of data scanned or improve performance. Option E is not correct because JSON is a row-based text format that is typically less efficient than Parquet for analytical queries and does not provide the columnar and compression benefits needed here.

Answer

Create views in Athena to simplify queries.

Answer

Convert the data to JSON format.

Question 7

Which TWO are common steps in exploratory data analysis?

Accepted Answer

Checking for missing values.

Answer

Training a machine learning model.

Answer

Deploying the model to production.

Answer

Tuning hyperparameters.

Question 8

A data scientist is training a deep learning model for image classification using Amazon SageMaker. The training job is taking too long. The data scientist wants to use distributed training across multiple GPUs to speed up the process. Which SageMaker feature should the data scientist use?

Accepted Answer

SageMaker Distributed Training Libraries. SageMaker Distributed Training Libraries provide optimized implementations of data parallelism and model parallelism that automatically partition the model and data across multiple GPUs, reducing training time for deep learning models. This is the correct choice because the question specifically asks for a feature to enable distributed training across multiple GPUs, which is exactly what these libraries are designed for.

Answer

SageMaker Managed Spot Training

Answer

SageMaker Hyperparameter Tuning

Answer

SageMaker Automatic Model Tuning

Question 9

A data scientist is exploring a dataset with many features and wants to detect multicollinearity. Which technique should the scientist use?

Accepted Answer

Calculate the Variance Inflation Factor (VIF) for each feature.. Variance Inflation Factor (VIF) is a standard metric for detecting multicollinearity. Option D (pairwise scatter plots) can hint at relationships but does not quantify multicollinearity. Option B (Pearson correlation matrix) shows pairwise linear correlation but does not capture multicollinearity among multiple variables. Option C (ANOVA) is used for comparing means, not for detecting multicollinearity.

Answer

Compute the Pearson correlation matrix between features.

Answer

Perform ANOVA on each feature against the target.

Answer

Create pairwise scatter plots of all features.

Question 10

During exploratory data analysis, a data scientist discovers that a feature has a variance of 0.01, while other features have variances around 1.0. Which action should be taken?

Accepted Answer

Consider removing the feature or applying variance threshold.. A feature with variance 0.01 is nearly constant, meaning it carries very little information for distinguishing between samples. In exploratory data analysis, such low-variance features are often candidates for removal because they contribute little to model performance and can add noise. Applying a variance threshold is a standard technique to filter out such features. Scaling or transforming would not increase its informational value.

Answer

Scale the feature to have unit variance.

Answer

Apply a log transformation to the feature.

Answer

Impute missing values in the feature.

Question 11

A company is building a recommendation system for an e-commerce platform. They have user-item interaction data and want to use matrix factorization. However, the dataset is sparse (99% missing interactions). Which approach should the data scientist take to train the model effectively?

Accepted Answer

Use alternating least squares (ALS) with implicit feedback and assign lower confidence to unobserved interactions. Alternating Least Squares (ALS) with implicit feedback is specifically designed to handle sparse implicit feedback datasets by assigning lower confidence to unobserved interactions (e.g., confidence = 1 + alpha * r_ui, where r_ui is 0 for unobserved). This avoids the pitfalls of treating missing values as zeros (which distorts the factorization) and scales well to 99% sparsity by leveraging weighted regularization.

Answer

Impute missing values with zeros and use singular value decomposition (SVD)

Answer

Remove all users and items with fewer than 10 interactions to reduce sparsity

Answer

Use item-based collaborative filtering with cosine similarity

Question 12

A data scientist is using Amazon SageMaker to train a model and wants to use a custom Docker container for training. The container requires access to a private Amazon ECR repository. Which IAM role configuration is needed?

Accepted Answer

Attach an IAM policy to the SageMaker execution role that allows ecr:GetDownloadUrlForLayer, ecr:BatchGetImage, and ecr:GetAuthorizationToken for the ECR repository.. When SageMaker runs a training job with a custom container from a private ECR repository, the training job assumes the SageMaker execution role. That role must include ECR permissions — specifically ecr:GetAuthorizationToken (to authenticate to ECR), plus ecr:BatchGetImage and ecr:GetDownloadUrlForLayer (to pull the image layers). Without these, the training job fails with an image pull error.

Answer

Use the AWS account owner's IAM role as the SageMaker execution role.

Answer

Create a new IAM user with ECR access and store credentials in SageMaker.

Answer

Add a bucket policy to the ECR repository allowing access from the SageMaker execution role.

Question 13

Which TWO actions can help reduce overfitting when training a model on SageMaker? (Choose TWO.)

Accepted Answer

Use early stopping based on validation error. Option B is correct because early stopping monitors a validation metric (e.g., validation error via the SageMaker EarlyStopping callback or the XGBoost early_stopping_rounds parameter) and halts training once the validation error stops improving, preventing the model from continuing to fit noise in the training set. Option E is correct because adding L1 regularization (in XGBoost, the alpha hyperparameter) penalizes large weights and drives some coefficients to zero, which constrains model complexity and reduces overfitting. Option A is wrong because reducing training data generally increases variance and makes overfitting worse, not better. Option C is wrong because increasing tree depth raises model capacity, allowing the model to memorize training data and overfit more. Option D is wrong because more training epochs let the model fit the training set even more closely, typically worsening overfitting.

Answer

Reduce the amount of training data

Answer

Increase the maximum depth of the trees in XGBoost

Answer

Increase the number of training epochs

Question 14

A company uses a linear regression model to predict house prices. The model's R-squared is 0.95 on the training set but 0.60 on the test set. Which of the following is the most likely cause?

Accepted Answer

Overfitting. A large gap between training R-squared (0.95) and test R-squared (0.60) is the classic signature of overfitting: the model has learned the training data too well, including noise, and fails to generalize to unseen data. Underfitting would show low performance on both sets, not a large disparity.

Answer

Underfitting

Answer

Multicollinearity among features

Answer

Data leakage

Question 15

A data scientist is training a linear regression model to predict house prices. The dataset includes features such as square footage, number of bedrooms, and location. After training, the model achieves an R² of 0.85 on the training set but only 0.60 on the test set. Which of the following is the MOST likely cause of this discrepancy?

Accepted Answer

The model is overfitting the training data. A high R² on the training set (0.85) paired with a significantly lower R² on the test set (0.60) is a classic symptom of overfitting. The model has learned noise and specific patterns in the training data that do not generalize to unseen data, causing poor test performance. Regularization techniques like Lasso or Ridge, or reducing model complexity, would typically address this issue.

Answer

There is multicollinearity among the features

Answer

The model is underfitting the training data

Answer

There is data leakage between the training and test sets