MLS-C01 Practice Test 38 — 15 Questions

Question 1

A data scientist is building a text classification model using a bag-of-words approach. The dataset contains 100,000 documents with a vocabulary of 50,000 unique words. The model is overfitting. Which THREE techniques can help reduce overfitting? (Choose THREE.)

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

Accepted Answer

Apply L1 or L2 regularization. Option B is correct because adding L1 or L2 regularization to the classifier penalizes large coefficients, shrinking the model's capacity and directly combating the high variance that causes overfitting on a 100,000-document bag-of-words matrix. Option C is correct because restricting the n-gram range to unigrams only removes higher-order bigram/trigram features, drastically cutting the feature space and reducing the model's ability to memorize spurious word co-occurrence patterns in the training set. Option D is correct because feature selection that removes rare words (e.g., dropping terms below a min_df threshold or using chi-squared/SelectKBest) eliminates noisy, low-frequency tokens that appear in few documents and are prime candidates for overfitting. Option A is not correct because increasing max_features expands the vocabulary and adds more parameters, which typically worsens overfitting rather than reducing it. Option E is not correct because switching from raw counts to TF-IDF is a weighting/normalization change that can improve performance but does not by itself constrain model complexity or reduce overfitting.

Answer

Increase max_features to include more words

Answer

Use TF-IDF instead of raw counts

Question 2

A data scientist is training a TensorFlow model on a single GPU instance. The training is taking too long. Which AWS service should be used to reduce training time by distributing the workload across multiple GPUs?

Accepted Answer

Amazon SageMaker. Amazon SageMaker provides built-in support for distributed training across multiple GPUs using its managed training infrastructure. By configuring a SageMaker training job with a 'distributed training' strategy (e.g., SageMaker's distributed data parallelism library), the TensorFlow model can automatically split the workload across multiple GPU instances, significantly reducing training time. SageMaker handles the underlying cluster orchestration, network setup, and fault tolerance, making it the correct choice for this scenario.

Answer

AWS Glue

Answer

Amazon EMR

Answer

AWS Batch

Question 3

A company uses Amazon SageMaker to train a linear regression model. During evaluation, they observe that the model has high bias (underfitting). Which THREE actions can reduce bias?

Accepted Answer

Add polynomial features.. Option B (Add polynomial features) is correct because high bias means the linear hypothesis class is too simple to capture the underlying relationship; adding polynomial features expands the feature space so the model can fit nonlinear patterns and reduce underfitting. Option C (Reduce the regularization strength) is correct because L2 regularization penalizes large weights and shrinks the model toward a simpler solution; lowering the regularization coefficient (for example, reducing alpha in the SageMaker LinearLearner predictor) allows the model more flexibility to fit the training data and lowers bias. Option E (Use a random forest model instead of linear regression) is correct because a random forest is a higher-capacity, non-linear ensemble that can represent complex decision boundaries a linear model cannot, thereby reducing bias. Option A is incorrect because increasing L2 regularization further constrains the model and increases bias, not reduces it. Option D is incorrect because using a smaller training dataset reduces the information available to the model, typically worsening underfitting rather than alleviating it.

Answer

Increase L2 regularization.

Answer

Use a smaller training dataset.

Question 4

Which TWO of the following are techniques used to reduce overfitting in a neural network?

Accepted Answer

L2 regularization. L2 regularization (C) is correct because it adds a penalty proportional to the sum of squared weights to the loss function, discouraging large weight values and thereby constraining model complexity to reduce overfitting. Dropout (D) is correct because it randomly deactivates a fraction of neurons during each training iteration, forcing the network to learn redundant, robust features and preventing co-adaptation that leads to overfitting. Increasing the number of layers (A) adds parameters and capacity, which typically worsens overfitting rather than reducing it. Batch normalization (B) primarily stabilizes and accelerates training by normalizing layer inputs; while it can have a mild regularizing side effect, it is not a dedicated overfitting-reduction technique. Increasing the learning rate (E) can cause unstable or divergent training and does not address overfitting.

Answer

Increase the number of layers

Answer

Batch normalization

Answer

Increase the learning rate

Question 5

A startup is using SageMaker to train a model using the built-in XGBoost algorithm. The training job runs successfully but the resulting model performs poorly on the test data. The data scientist suspects overfitting. The training data is relatively small (10,000 rows). Which action should be taken to reduce overfitting?

Accepted Answer

Decrease the number of trees (num_round) to 50. Reducing the number of boosting rounds (num_round) decreases model complexity, which is a direct way to combat overfitting in XGBoost when the training set is small. With fewer trees, the model has less capacity to memorize noise in the 10,000-row dataset. This is a standard regularization technique for gradient boosting.

Answer

Increase the learning rate to 0.3

Answer

Increase the number of trees (num_round) to 500

Answer

Use a larger instance type

Question 6

Which THREE techniques are commonly used to detect multicollinearity in a dataset during exploratory data analysis?

Accepted Answer

Eigenvalue analysis from PCA. Eigenvalue analysis from PCA (B) is correct because multicollinearity manifests as very small eigenvalues in the correlation/covariance matrix, indicating that some linear combinations of predictors explain almost no variance. A correlation matrix (C) is correct because high pairwise correlation coefficients (e.g., |r| > 0.8 or 0.9) between independent variables are a direct, common indicator of multicollinearity. Variance Inflation Factor (VIF) (D) is correct because VIF quantifies how much the variance of an estimated regression coefficient is inflated by collinearity, with values above 5 or 10 typically signaling problematic multicollinearity. A heatmap of missing values (A) only visualizes nullity patterns and says nothing about linear dependence among features, and a scatter matrix of all features (E) is a general pairwise visualization that may hint at relationships but is not a standard quantitative multicollinearity detection technique like VIF, correlation coefficients, or PCA eigenvalues.

Answer

Heatmap of missing values

Answer

Scatter matrix of all features

Question 7

Which TWO of the following are valid approaches to handle missing values in a dataset for a machine learning model?

Accepted Answer

Impute missing values with the mean of the column. Option B is correct because mean imputation replaces missing numeric entries with the column's average, a standard and simple technique that preserves the row count and avoids discarding data. Option C is correct because listwise deletion (dropping rows containing missing values) is a valid, widely used approach when the missingness is random and the remaining data is sufficient. Option A is not a standard or reliable missing-value handling method; neural networks can model data but are not a defined imputation technique and would themselves require complete training data. Option D is incorrect because standardization rescales feature values and does nothing to fill or remove missing entries. Option E is incorrect because one-hot encoding is for converting categorical variables into binary vectors, not for handling missing values.

Answer

Use a neural network to predict missing values

Answer

Standardize the features to handle missing values

Answer

Apply one-hot encoding to convert missing values

Question 8

A company wants to use Amazon SageMaker to train a model using a custom algorithm packaged in a Docker container. Which approach should they use?

Accepted Answer

Use the SageMaker SDK to create an Estimator with the image URI of the custom container. The correct approach is to use the SageMaker SDK to create an Estimator with the image URI of the custom container, as SageMaker supports bring-your-own-container for custom algorithms. Option A is incorrect because SageMaker Ground Truth is a labeling service, not for training custom algorithms. Option B is incorrect because SageMaker Autopilot automates model selection and tuning, but it does not support custom containers. Option D is incorrect because built-in algorithms are predefined and do not allow custom code.

Answer

Use SageMaker Ground Truth

Answer

Use SageMaker Autopilot

Answer

Select one of the built-in algorithms in SageMaker

Question 9

A machine learning team is building a recommendation system for an e-commerce platform. They have user-item interaction data (clicks, purchases). They need to choose an algorithm that can capture both user and item latent factors and handle missing data. Which algorithm should they use?

Accepted Answer

Matrix factorization. Matrix factorization is the correct choice because it decomposes the user-item interaction matrix into lower-dimensional latent factors for users and items, capturing underlying patterns in preferences. It naturally handles missing data by learning from observed interactions only, making it ideal for recommendation systems with sparse data.

Answer

Linear regression

Answer

Principal component analysis (PCA)

Answer

Convolutional neural network (CNN)

Question 10

A team is training a deep learning model using TensorFlow on a single GPU instance in SageMaker. The GPU utilization is below 30%. Which change will MOST improve GPU utilization?

Accepted Answer

Increase the batch size. Low GPU utilization (below 30%) usually means the GPU is starved for work because the batch size is too small to saturate its compute units, so increasing the batch size gives the GPU more data to process per step and raises utilization. This is the most direct, single-instance change that improves GPU efficiency without altering the training architecture or requiring distributed infrastructure. Reducing epochs or switching to CPU does not address utilization, and distributed training adds complexity without fixing the underlying per-GPU starvation.

Answer

Reduce the number of epochs

Answer

Use SageMaker Distributed Training with multiple GPUs

Answer

Switch to a CPU instance

Question 11

A company uses Amazon EMR with Spark to process data daily. The job reads from S3 and writes to S3. Recently, the job started failing with 'S3AccessDenied' errors. The IAM role used by EMR has not changed. What is the MOST likely cause?

Accepted Answer

The S3 bucket policy was updated to deny access to the EMR role. The IAM role used by EMR has not changed, but the S3 bucket policy can be updated independently to deny access to that specific role. An explicit deny in a bucket policy overrides any allow in the IAM policy, causing the 'S3AccessDenied' error even though the role itself remains unchanged. This is the most likely cause because the job was previously working and only the bucket policy could have been modified without touching the EMR configuration.

Answer

The EMR cluster's security group blocks outbound traffic

Answer

The S3 bucket was deleted and recreated

Answer

The EMR service role was rotated

Question 12

A data scientist is starting a new machine learning project and needs to understand the dataset. The dataset is stored as CSV files in Amazon S3, with a total size of 50 GB. The data scientist wants to quickly get summary statistics (count, mean, standard deviation, min, max) for each numerical column, and also check for missing values. The data scientist has access to SageMaker Studio. What is the most efficient way to achieve this?

Accepted Answer

Use SageMaker Data Wrangler to import the data and generate a data quality report.. SageMaker Data Wrangler is purpose-built for data preparation and profiling, allowing you to compute summary statistics and check for missing values with a visual interface and without writing code. Option A (AWS Glue Crawler + Athena) only infers schema and enables SQL queries; it does not automatically provide summary statistics or missing value counts. Option B (PySpark script) is possible but requires manual coding and Spark cluster management, making it less efficient for quick exploration. Option C (Amazon QuickSight) is a BI tool that requires loading data into SPICE, which is not as streamlined for initial data profiling as Data Wrangler.

Answer

Use AWS Glue Crawler to infer schema and then query with Athena.

Answer

Write a PySpark script in a SageMaker notebook to compute statistics.

Answer

Load a sample into Amazon QuickSight and use SPICE to compute statistics.

Question 13

A company is building a real-time fraud detection system using Amazon SageMaker. The model must have low latency (under 10ms) and high throughput (thousands of predictions per second). The team has trained a gradient boosting model using XGBoost. Which SageMaker inference option is MOST suitable?

Accepted Answer

Deploy the model on a SageMaker real-time endpoint with a multi-model endpoint.. A multi-model endpoint (MME) on SageMaker is the most suitable option because it allows you to host multiple XGBoost models on a single endpoint, sharing the underlying instance to maximize throughput and minimize latency. MMEs keep models loaded in memory and route requests to the correct model with sub-10ms overhead, meeting the low-latency and high-throughput requirements for real-time fraud detection.

Answer

Use SageMaker asynchronous inference.

Answer

Deploy the model on a SageMaker serverless endpoint.

Answer

Use a batch transform job.

Question 14

A data engineer is building a data pipeline that ingests streaming data from IoT devices. The data must be processed in near real-time and stored in Amazon S3 for further analysis. Which AWS service should be used to capture and process the streaming data before storing it in S3?

Accepted Answer

Use Amazon Kinesis Data Streams to capture the data and Amazon Kinesis Data Firehose to deliver it to S3.. Amazon Kinesis Data Streams is designed for real-time data ingestion and can capture streaming data from IoT devices with low latency. Amazon Kinesis Data Firehose then reliably loads that streaming data into Amazon S3, handling buffering, compression, and partitioning automatically. This combination provides the near real-time processing and durable storage required for the pipeline.

Answer

Use Amazon S3 with S3 Event Notifications to trigger AWS Lambda for processing.

Answer

Use AWS Glue to perform ETL on the streaming data.

Answer

Use Amazon Simple Queue Service (SQS) to buffer the data and then process it with AWS Lambda.

Question 15

A data scientist wants to explore a large dataset stored in Amazon S3 using SQL queries without moving the data. The dataset is in CSV format and is updated daily with new partitions. Which AWS service should be used to directly query the data in S3?

Accepted Answer

Amazon Athena. Amazon Athena is a serverless interactive query service that allows you to analyze data directly in Amazon S3 using standard SQL, without needing to load or transform the data. It supports CSV format and can automatically discover new partitions when used with Hive-style partitioning and the MSCK REPAIR TABLE command or by enabling partition projection. This makes it the ideal choice for directly querying a large, daily-updated CSV dataset stored in S3.

Answer

Amazon Redshift Spectrum

Answer

Amazon EMR

Answer

AWS Glue