MLS-C01 Practice Test 46 — 15 Questions

Question 1

A data scientist is training a linear regression model on a dataset with 10 numerical features. After training, the model's R-squared value is 0.99 on the training set but only 0.60 on the test set. Which TWO of the following are appropriate actions to reduce overfitting? (Choose TWO.)

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

Accepted Answer

Use a subset of the most important features. Option C is correct because using a subset of the most important features (feature selection) reduces model complexity and variance, which directly combats overfitting when the model has learned noise from irrelevant or redundant features. Option E is correct because L2 regularization (Ridge regression) adds a penalty term equal to lambda times the sum of squared coefficients to the loss function, shrinking coefficients and constraining model complexity to improve generalization. Option A is not correct because normalizing features only rescales inputs to comparable ranges; it aids convergence and fair regularization but does not by itself reduce overfitting. Option B is not correct because adding more features increases model complexity and typically worsens overfitting. Option D is not correct because increasing training epochs lets the model fit the training data even more closely, worsening the train-test gap.

Answer

Normalize the features

Answer

Add more features to the model

Answer

Increase the number of training epochs

Question 2

A data scientist is analyzing a dataset with high cardinality categorical features (e.g., user IDs with millions of unique values). They want to visualize the relationship between these categorical features and a continuous target variable. Which approach is most effective for EDA?

Accepted Answer

Group rare categories into an 'Other' category and use box plots. For high-cardinality categorical features, plotting every category is impractical. Grouping rare categories into an 'Other' bucket reduces the number of categories to a manageable set, and box plots then show the distribution of the continuous target variable across each category, revealing median, spread, and outliers. This is the most effective EDA approach because it balances information retention with visual clarity.

Answer

Apply one-hot encoding and use scatter plots

Answer

Use a bar chart with all categories on x-axis

Answer

Remove the categorical features from analysis

Answer

Apply feature hashing and visualize the hashed values

Question 3

A data scientist is using Amazon SageMaker to train a model. The training data is stored in an S3 bucket encrypted with AWS KMS. Which TWO actions are necessary to allow SageMaker to access the data?

Accepted Answer

Ensure the SageMaker execution role has s3:GetObject permission.. A is correct because the SageMaker execution role must have the s3:GetObject permission to read objects from the S3 bucket. Without this IAM permission, SageMaker cannot retrieve the training data, even if the bucket is otherwise accessible.

Answer

Enable S3 Transfer Acceleration.

Answer

Set up a VPC endpoint for S3.

Answer

Add a bucket policy allowing SageMaker access.

Question 4

Which TWO statements about handling missing data during EDA are correct? (Select TWO.)

Accepted Answer

If data are missing completely at random (MCAR), listwise deletion yields unbiased estimates.. Options C and E are correct. Option C is correct because when data are Missing Completely at Random (MCAR), the missingness is independent of both observed and unobserved data, so listwise deletion (removing rows with missing values) does not introduce bias; the remaining sample is still a random subsample. Option E is correct because the median is not influenced by extreme values, making it a more robust imputation method compared to the mean, which can be skewed by outliers. Option A is incorrect because dropping columns with >50% missing values is not always recommended; it depends on the importance of the variable and the analysis goals. Option B is incorrect because mean imputation reduces the variance of the imputed variable, as it forces imputed values to the center. Option D is incorrect because Multiple Imputation by Chained Equations (MICE) is not always the safest; it assumes data are Missing at Random (MAR) and can be complex or inappropriate for other missingness mechanisms.

Answer

Dropping columns with >50% missing values is always recommended.

Answer

Mean imputation preserves the variance of the original distribution.

Answer

Multiple imputation (MICE) is always the safest method regardless of missing data mechanism.

Question 5

A company wants to monitor SageMaker endpoints for data drift. Which TWO services can be used together to detect and alert on drift?

Accepted Answer

SageMaker Model Monitor. SageMaker Model Monitor (option B) is the AWS service purpose-built to continuously monitor SageMaker endpoints for data drift, model quality, bias, and feature attribution drift; it captures endpoint traffic, compares it against a baseline, and emits metrics to CloudWatch. Amazon CloudWatch Alarms (option D) is then used to watch those Model Monitor metrics (e.g., the drift metric emitted to the AWS/SageMaker namespace) and trigger notifications or actions when drift exceeds a threshold, which is exactly the detect-and-alert workflow the company needs. SageMaker Data Wrangler (option A) is a data preparation and feature-engineering tool, not a runtime drift-detection service, so it does not monitor live endpoints. AWS CodePipeline (option C) is a CI/CD orchestration service and has no role in detecting data drift. Amazon CloudWatch Logs (option E) stores log data but does not itself detect drift or raise alerts, so it cannot fulfill the alerting requirement on its own.

Answer

SageMaker Data Wrangler

Answer

AWS CodePipeline

Answer

Amazon CloudWatch Logs

Question 6

A company is using Amazon SageMaker to train a XGBoost model for predicting customer churn. The training data is stored in an S3 bucket as CSV files. The data scientist runs a hyperparameter tuning job with 50 training jobs. The tuning job completes, but the best model's accuracy on the holdout set is lower than expected. The data scientist suspects that the hyperparameter ranges are too narrow. Which corrective action is most appropriate?

Accepted Answer

Expand the hyperparameter ranges for key parameters such as 'max_depth', 'learning_rate', and 'subsample'. The data scientist suspects the hyperparameter ranges are too narrow, which directly limits the model's ability to find an optimal configuration. Expanding ranges for key XGBoost parameters like 'max_depth', 'learning_rate', and 'subsample' allows the tuning job to explore a broader space of model complexities and regularization levels, potentially improving accuracy on the holdout set. This is the most direct fix for the stated problem, as it addresses the root cause rather than increasing job count or changing the search strategy.

Answer

Increase the number of training jobs in the tuning job

Answer

Switch to a different algorithm like Random Forest

Answer

Change the tuning strategy from random search to Bayesian optimization

Question 7

A machine learning engineer is deploying a model using Amazon SageMaker and wants to automatically scale the endpoint based on the number of incoming requests. Which scaling policy should be used?

Accepted Answer

Target tracking scaling. Target tracking scaling is correct because it automatically adjusts the number of endpoint instances to keep a specified metric (like InvocationsPerInstance) at a target value, directly responding to incoming request load. SageMaker's target tracking policy uses a predefined or custom metric and a target value, then scales in or out to maintain that target, which is ideal for request-based scaling. This is the only policy among the options that continuously adjusts capacity based on real-time demand without manual threshold definitions.

Answer

Step scaling

Answer

Scheduled scaling

Answer

Simple scaling

Question 8

A data scientist is using Amazon SageMaker to train a model with the built-in DeepAR algorithm for time series forecasting. The dataset contains 100 related time series, each with 2 years of daily data. The data scientist notices that the model performs well on most series but poorly on a few series with intermittent demand (many zero values). Which modification to the training data or algorithm configuration is most likely to improve performance on the intermittent series?

Accepted Answer

Set the likelihood parameter to 'negative-binomial' to better model count data with zeros.. Intermittent demand series are characterized by many zeros and occasional positive counts. DeepAR's default likelihood may not handle this well. Switching to the negative binomial likelihood allows the model to account for overdispersion and zero inflation, leading to better forecasts for these series. This is a targeted configuration change that addresses the underlying data distribution.

Answer

Increase the number of epochs to allow the model to better learn the zero patterns.

Answer

Remove the intermittent series from the training dataset to prevent them from degrading overall model performance.

Answer

Increase the context_length hyperparameter to include more historical data.

Question 9

A data scientist is deploying a model on Amazon SageMaker for real-time inference. The model is a PyTorch model that requires custom inference code. The data scientist needs to handle variable-length inputs and optimize inference latency. Which TWO steps should the data scientist take? (Choose TWO.)

Accepted Answer

Use TorchScript to compile the model for optimized inference.. Option D is correct because TorchScript compiles the PyTorch model into a serialized, optimized representation that can be executed independently of Python, reducing per-request overhead and improving real-time inference latency, which directly addresses the optimization requirement. Option E is correct because SageMaker's PyTorch inference toolkit requires a custom inference.py script that implements model_fn to load the model and input_fn, predict_fn, and output_fn (or transform_fn) to handle variable-length inputs and custom preprocessing/postprocessing for real-time endpoints. Option A is incorrect because batch transform is an offline, asynchronous batch-processing feature, not a real-time inference endpoint, so it does not satisfy the real-time latency scenario. Option B is incorrect because an unmodified PyTorch container lacks the custom inference code needed to load the model and handle variable-length inputs. Option C is incorrect because multiple variants support A/B testing traffic splits, not latency optimization or variable-length input handling.

Answer

Enable SageMaker batch transform to process requests in batches.

Answer

Use the SageMaker PyTorch container without any modifications.

Answer

Set the endpoint to use multiple variants for A/B testing.

Question 10

A data scientist is training a deep learning model using Amazon SageMaker. The training loss is decreasing, but the validation loss starts increasing after 10 epochs. The model is overfitting. Which TWO actions should the data scientist take to reduce overfitting? (Choose 2.)

Accepted Answer

Add dropout layers. Option D is correct because adding dropout layers randomly deactivates a fraction of neurons during training, which forces the network to learn more robust, distributed features and is a standard regularization technique to combat overfitting. Option E is correct because early stopping monitors validation loss and halts training once it stops improving (e.g., after the 10-epoch point where validation loss begins rising), preventing the model from continuing to memorize the training data. Option A is incorrect because increasing the number of layers raises model capacity, which would worsen overfitting rather than reduce it. Option B is incorrect because removing L2 regularization eliminates a penalty on large weights, reducing regularization and increasing overfitting. Option C is incorrect because increasing training steps lets the model fit the training data even longer, further degrading validation performance.

Answer

Increase the number of layers

Answer

Remove L2 regularization

Answer

Increase the number of training steps

Question 11

A data scientist is performing EDA on a time series dataset of daily website visits. The scientist wants to identify any seasonality patterns. Which visualization is most appropriate?

Accepted Answer

Line plot with day on x-axis and visits on y-axis, highlighting weekends.. A line plot with day on the x-axis and visits on the y-axis, with weekends highlighted, can reveal weekly seasonality patterns. Option A (correlation matrix with lags) can detect autocorrelation but is not a direct visualization of seasonality. Option B (scatter plot vs day of month) could show monthly patterns but is less effective for daily seasonality and does not preserve time order as clearly as a line plot. Option C (histogram) shows distribution, not temporal patterns. Therefore, option D is best.

Answer

Correlation matrix of visits with lagged versions of itself.

Answer

Scatter plot of visits against the day of the month.

Answer

Histogram of daily visit counts.

Question 12

A data engineering team needs to process streaming data from thousands of IoT devices. They want to aggregate data in 1-minute windows and store results in an S3 data lake for downstream analytics. Which architecture should they use?

Accepted Answer

Use Kinesis Data Analytics for SQL-based windowed aggregations and send results to Kinesis Data Firehose for delivery to S3.. Kinesis Data Analytics for SQL Applications is purpose-built for real-time windowed aggregations on streaming data, such as 1-minute tumbling windows. It can directly consume from Kinesis Data Streams, perform the aggregation using standard SQL, and output the results to Kinesis Data Firehose, which reliably delivers the aggregated data to an S3 data lake with built-in buffering and compression.

Answer

Use AWS Glue ETL jobs running in streaming mode to read from Kinesis Data Streams, apply window aggregations, and write to S3.

Answer

Use Kinesis Data Streams with enhanced fan-out and multiple consumers to aggregate windows, then write to S3 via Firehose.

Answer

Use Kinesis Data Streams, trigger a Lambda function for 1-minute window aggregation using Python, and write results to S3.

Question 13

A data scientist is analyzing a dataset with 1,000 features. They suspect many features are redundant and want to reduce dimensionality before training a model. Which technique is most appropriate for identifying the most important features?

Accepted Answer

Use L1 regularization (Lasso) to shrink coefficients to zero. L1 regularization (Lasso) adds a penalty equal to the absolute value of the coefficients to the loss function, which drives less important feature coefficients exactly to zero. This makes Lasso inherently a feature-selection technique: the features with non-zero coefficients after training are the ones the model retained as important, directly addressing the goal of identifying the most important features among 1,000.

Answer

Apply principal component analysis (PCA) and select the top components

Answer

Train a random forest and remove features with low importance

Answer

Compute the correlation matrix and remove features with high correlation

Question 14

A data scientist is performing EDA on a dataset with mixed data types (numerical and categorical). Which TWO visualizations are most appropriate for understanding the distribution of categorical features?

Accepted Answer

Pie chart. A pie chart (C) is appropriate because it displays the proportion or percentage share of each category relative to the whole, making it easy to see how a categorical feature's levels are distributed when the number of categories is small. A bar chart (E) is also appropriate because it plots the frequency or count of each category as separate bars, clearly revealing the distribution, dominance, and relative sizes of categorical values. Histograms (A) are designed for continuous numerical data by binning values, so they do not represent discrete categories meaningfully. Box plots (B) summarize numerical distributions via quartiles, median, and outliers, which is not applicable to categorical data. Scatter plots (D) show relationships between two numerical variables and cannot meaningfully plot categorical distributions.

Answer

Histogram

Answer

Box plot

Answer

Scatter plot

Question 15

A team has a dataset with 500 features and wants to reduce dimensionality. During EDA, they compute the variance of each feature. Which finding would most likely lead to feature removal?

Accepted Answer

Some features have near-zero variance. Features with near-zero variance carry almost no information — they are nearly constant across all observations, so they contribute little to distinguishing samples or predicting the target. Removing them reduces dimensionality with minimal information loss. This is the standard low-variance filter in feature selection.

Answer

Some features have high correlation with each other

Answer

Some features have negative covariance with the target

Answer

Some features have very high variance