MLS-C01 Practice Exam 62 — 20 Questions

Question 1

A data analyst is performing exploratory data analysis on a dataset with 100 features. The analyst wants to identify which features contribute most to the variance in the data. Which technique should the analyst use?

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

Accepted Answer

Principal Component Analysis (PCA). Principal Component Analysis (PCA) is the correct technique because it is an unsupervised dimensionality reduction method that identifies the principal components, which are linear combinations of the original features that capture the maximum variance in the data. Option A (K-means) is incorrect because it is a clustering algorithm that groups data points, not used for analyzing feature variance. Option C (t-SNE) is incorrect because it is primarily used for visualizing high-dimensional data in lower dimensions but does not provide explicit variance contributions. Option D (LDA) is incorrect because it is a supervised method that requires class labels and aims to maximize class separability, not variance.

Answer

K-means clustering

Answer

t-Distributed Stochastic Neighbor Embedding (t-SNE)

Answer

Linear Discriminant Analysis (LDA)

Question 2

A data scientist is using Amazon SageMaker to train a model with the built-in XGBoost algorithm. The dataset contains missing values. What is the default behavior of SageMaker XGBoost regarding missing values?

Accepted Answer

It automatically learns the best direction (left or right) for missing values during training. SageMaker's built-in XGBoost algorithm treats missing values (i.e., NaN, None, or 0 by default) as a separate category. During training, it automatically learns the best direction (left or right) to assign missing values at each split, based on the reduction in loss. This is the default behavior (Option D). Option A is incorrect because XGBoost does not raise an error; it handles missing values internally. Option B is incorrect because XGBoost does not impute with the column mean; that would require data preprocessing. Option C is incorrect because rows with missing values are not removed; they are included in training with the learned direction.

Answer

It raises an error and stops training

Answer

It imputes missing values with the column mean

Answer

It removes rows with missing values

Question 3

A company is using Amazon SageMaker to train a deep learning model. The training job uses a script that reads data from Amazon S3 using the SageMaker SDK's `s3_input` method. The training job runs on a single ml.p3.2xlarge instance. The data scientist notices that the GPU utilization is very low during training, often below 20%. The training dataset is large, approximately 50 GB, stored as TFRecord files in S3. What is the MOST likely cause of low GPU utilization?

Accepted Answer

The data loading pipeline is not optimized, causing the GPU to wait for data.. Low GPU utilization typically indicates that the GPU is waiting for data, which is a classic symptom of a data loading bottleneck. With a large 50 GB TFRecord dataset and a single ml.p3.2xlarge instance, the default SageMaker SDK's s3_input method may not be optimized for high throughput. To fully utilize the GPU, techniques such as using Pipe mode, prefetching, parallel data extraction, and using a data loader like TensorFlow's tf.data API with interleave and prefetch are recommended. Option B correctly identifies this bottleneck. Option A is incorrect because a CPU-only version of TensorFlow would not run on a GPU at all. Option C is incorrect because a batch size that is too large causes out-of-memory errors, not low utilization. Option D is incorrect because ml.p3.2xlarge has 8 GB of GPU memory, which is suitable for many models; low memory would cause failures rather than low utilization.

Answer

The training script is using a CPU-only version of TensorFlow.

Answer

The batch size is too large, causing the GPU to run out of memory.

Answer

The instance type does not have enough GPU memory for the model.

Question 4

A data scientist is using Amazon SageMaker to train a model using a built-in algorithm. The training job fails with an error indicating that the algorithm expects the data to be in recordIO-protobuf format, but the input is CSV. What is the most efficient way to resolve this?

Accepted Answer

Use a SageMaker processing job to convert the CSV data to recordIO-protobuf format.. Using a SageMaker processing job to convert CSV data to recordIO-protobuf format is an efficient, scalable, and fully managed solution within the SageMaker ecosystem. Option A is incorrect because changing the inference data to recordIO-protobuf does not address the training data format issue. Option B is incorrect because using a boto3 script to convert locally is less efficient and not scalable compared to a managed processing job. Option D is incorrect because switching algorithms may not be desirable and does not solve the underlying data format requirement.

Answer

Change the inference data to recordIO-protobuf format.

Answer

Use a boto3 script to convert the CSV files locally and upload.

Answer

Switch to a different algorithm that accepts CSV format.

Question 5

A data scientist is training a text classification model using Amazon SageMaker's BlazingText algorithm. The dataset consists of 1 million documents, each labeled with one of 10 categories. The model achieves 92% accuracy on a held-out test set. However, when deployed, the model performs poorly on documents containing slang and typos. What should the data scientist do to improve model robustness?

Accepted Answer

Augment the training data by introducing common slang replacements and typos.. Data augmentation by introducing common slang replacements and typos into the training data increases the model's robustness to such variations, helping it generalize better to real-world text that contains slang and typos. Removing such documents (Option A) reduces the training data and does not teach the model to handle these variations. Increasing the embedding dimension (Option C) or number of epochs (Option D) does not directly address the issue of slang and typos.

Answer

Remove all documents with slang or typos from the training set.

Answer

Increase the embedding dimension from 100 to 300.

Answer

Increase the number of training epochs.

Question 6

A team is building a regression model to predict house prices. The dataset includes a column 'zip_code' with 100 unique values. The data scientist one-hot encodes this column, resulting in 100 new binary columns. The model shows poor performance on a validation set. What is the most likely cause?

Accepted Answer

The model suffers from the curse of dimensionality due to the large number of features.. One-hot encoding 'zip_code' with 100 unique values creates 100 binary features. When combined with other features, the total number of features can be large relative to the number of training samples, leading to the curse of dimensionality. This causes the model to overfit the training data and generalize poorly to the validation set. While multicollinearity among one-hot encoded columns is often low due to their binary nature, the primary issue here is the high dimensionality relative to sample size. Option D correctly identifies this as the most likely cause.

Answer

One-hot encoding introduced multicollinearity among the binary columns.

Answer

One-hot encoding reduced the number of features, causing underfitting.

Answer

The one-hot encoding introduced high variance, but the validation set has low variance.

Question 7

A company runs a data pipeline using AWS Glue ETL jobs that process about 10 TB of data daily from Amazon S3. The jobs are triggered by a schedule and write results to a separate S3 bucket. Recently, the jobs have been taking longer to complete, and the data engineering team has observed that the number of files in the source bucket has increased significantly, from thousands to millions of small files (each about 100 KB). The Glue jobs are configured to use the 'Group Files' option, but performance is still poor. The team needs to improve the job performance without changing the source data generation process. Which course of action should the team take?

Accepted Answer

Create a separate Glue job that runs before the main job to consolidate small files into larger ones in the source bucket. The main performance bottleneck is the large number of small files, which causes high overhead in reading metadata and opening files. Option D addresses this by creating a separate Glue job that consolidates small files into larger files (e.g., 100 MB) before the main ETL job runs, reducing the file count and improving read performance. Option A is incorrect because increasing DPUs may provide more parallelism but does not solve the underlying small-file problem; the overhead of opening millions of files remains. Option B is incorrect because switching to Amazon EMR with Spark would still encounter the same small-file issue unless additional measures (like coalesce or file compaction) are taken, which Option D already provides. Option C is incorrect because AWS Lambda has limitations on execution duration and memory, making it impractical to pre-process millions of small files efficiently.

Answer

Increase the number of DPUs allocated to the existing Glue job

Answer

Switch the ETL processing to Amazon EMR with Spark

Answer

Use AWS Lambda to pre-process the files and combine them

Question 8

A company is building a fraud detection model using a random forest classifier. The dataset is highly imbalanced with 99% legitimate transactions and 1% fraudulent. The model currently achieves 99% accuracy on the test set, but the fraud recall is only 10%. The business requires at least 80% recall for fraud. The data scientist has tried oversampling the minority class and adjusting class weights, but recall remains below 40%. The dataset contains millions of transactions with hundreds of features. Which approach should the data scientist try next to improve fraud recall?

Accepted Answer

Use a gradient boosting machine (e.g., XGBoost) with scale_pos_weight parameter. The current model uses random forest but still has low recall after oversampling and class weights. Gradient boosting machines like XGBoost can handle imbalanced data well with the scale_pos_weight parameter, which adjusts the weight of the positive class. This approach often yields better recall than random forest on very imbalanced datasets. Option A (random undersampling) would discard too much majority data and reduce model performance. Option C (PCA) is not directly targeting the imbalance issue and may remove important features. Option D (logistic regression) is a linear model that typically underperforms on complex, high-dimensional data with severe imbalance.

Answer

Randomly undersample the majority class to a 50:50 ratio

Answer

Apply PCA to reduce dimensionality before training

Answer

Use a logistic regression model with L2 regularization

Question 9

A company uses Amazon SageMaker to host a model for real-time inference. The model is a large ensemble that takes 2 seconds to load into memory. To reduce cold start latency, the data scientist uses SageMaker's managed warm pools. However, they notice that during a sudden traffic spike, new instances still experience high latency. What is the BEST way to ensure consistently low latency for all requests?

Accepted Answer

Configure auto scaling based on the number of active invocations to maintain a buffer of warmed instances.. Configuring auto scaling based on the number of active invocations maintains a buffer of warmed instances. This ensures that when traffic spikes occur, new instances are already loaded and ready to serve requests, avoiding cold start latency. Option A is wrong because using a larger instance type does not eliminate cold starts; the model still needs to load into memory. Option C is wrong because reducing instances increases the frequency of cold starts. Option D is wrong because SageMaker Serverless Inference has its own cold start overhead and is not suitable for workloads requiring consistently low latency.

Answer

Use a larger instance type to reduce model loading time.

Answer

Reduce the number of instances to minimize cold start frequency.

Answer

Switch to SageMaker Serverless Inference.

Question 10

A data scientist is using Amazon SageMaker Autopilot to automatically build a binary classification model. The dataset has 50 features and 100,000 rows. After the experiment completes, the best candidate model achieves an F1 score of 0.85 on the validation set. However, when deployed to a real-time endpoint, the model's F1 score drops to 0.72 on production data. The data distributions between training and production are similar. What is the MOST likely cause of the performance drop?

Accepted Answer

The Autopilot pipeline used features that are not available at inference time (data leakage).. Data leakage during Autopilot's feature engineering can lead to overly optimistic validation scores because features may have been derived from the target variable or from future information. Option A is wrong because similar distributions suggest no drift. Option B is wrong because Autopilot handles missing values. Option C is irrelevant because inference instance type does not affect model accuracy.

Answer

Concept drift occurred between training and production.

Answer

The production data contains missing values that were not present in training.

Answer

The inference endpoint uses a different instance type than training.

Question 11

A company is using Amazon SageMaker to train a linear learner model for predicting customer lifetime value. The target variable is right-skewed with a long tail. The data scientist applies a log transformation to the target variable and trains the model. The model achieves a low root mean squared error (RMSE) on the log scale. However, when the predictions are exponentiated back to the original scale, the RMSE is much higher. Which step should the data scientist take to improve the model's performance on the original scale?

Accepted Answer

Use a loss function that models the original distribution, such as Poisson or Tweedie. Using a loss function like Poisson or Tweedie directly models the non-negative, skewed distribution of the target variable in its original scale, which avoids the bias introduced by log transformation when predicting on the original scale. Option A (increase regularization) may not address the scale mismatch. Option B (remove outliers) could discard valuable data. Option D (use a deep learning model) might not solve the fundamental issue of loss function selection.

Answer

Increase the regularization strength

Answer

Remove outliers from the training data

Answer

Use a deep learning model instead of linear learner

Question 12

A research team is training a deep learning model for image classification using Amazon SageMaker. The model is a convolutional neural network (CNN) with 50 layers. The team uses a single ml.p3.2xlarge instance. After 10 hours of training, the model has not converged and the loss is decreasing very slowly. The team suspects vanishing gradients. They want to diagnose and fix the issue without significant code changes. Which action should they take?

Accepted Answer

Modify the architecture to include residual connections (skip connections). (Modify the architecture to include residual connections) directly addresses vanishing gradients by allowing gradients to flow through skip connections. Option A (adding more layers) worsens the problem. Option C (batch normalization) helps but is not as targeted as residual connections. Option D (increase learning rate) may cause divergence.

Answer

Add more convolutional layers to increase model capacity

Answer

Use batch normalization after each convolutional layer

Answer

Increase the learning rate by a factor of 10

Question 13

A data scientist is using Amazon SageMaker to train a linear learner model for regression. After reviewing the training logs, the data scientist notices that the loss is not decreasing and remains high. The learning rate is set to 0.01. The data is normalized. What should the data scientist do to improve convergence?

Accepted Answer

Try different learning rates, such as 0.001 or 0.1.. The loss is not decreasing, which often indicates an inappropriate learning rate. Trying different learning rates (e.g., 0.001 or 0.1) can help find a rate that allows the model to converge. Option A is incorrect because the data is already normalized; normalizing again would have no effect. Option B is incorrect because reducing the mini-batch size introduces more stochasticity but does not directly address learning rate issues. Option D is incorrect because increasing epochs does not help if the learning rate prevents convergence; the model may continue to oscillate or plateau.

Answer

Normalize the data again.

Answer

Reduce the mini-batch size.

Answer

Increase the number of epochs.

Question 14

Which TWO services can be used to serve machine learning models for real-time inference? (Select TWO.)

Accepted Answer

Amazon SageMaker. Amazon SageMaker provides real-time endpoints. AWS Lambda can be used with a custom container for inference. Amazon Rekognition is a specialized service for image analysis, not general model serving. Amazon Comprehend is NLP service. Amazon Athena is query service.

Answer

Amazon Rekognition

Answer

Amazon Athena

Answer

Amazon Comprehend

Question 15

A data scientist is training a recurrent neural network (RNN) for time series forecasting. The model's training loss is not decreasing, and the gradients are vanishing. Which technique should the scientist apply to address vanishing gradients?

Accepted Answer

Replace the RNN cells with LSTM or GRU units.. LSTM and GRU units incorporate gating mechanisms (input, forget, output gates) that regulate the flow of gradients, effectively mitigating the vanishing gradient problem. Option A is wrong because gradient clipping is used to prevent exploding gradients, not vanishing gradients. Option C is wrong because batch normalization helps stabilize training by reducing internal covariate shift, but it does not specifically address vanishing gradients. Option D is wrong because increasing the learning rate may cause the training to diverge or become unstable, and it does not solve the vanishing gradient issue.

Answer

Apply gradient clipping.

Answer

Add batch normalization layers.

Answer

Increase the learning rate.

Question 16

A data scientist is working with a dataset containing customer transaction records stored in Amazon S3 as CSV files. The dataset has 500 columns and 2 million rows. The scientist wants to perform EDA to understand data types, missing values, and summary statistics for each column. They need to do this quickly and without writing custom code. The scientist has access to AWS Glue DataBrew and Amazon SageMaker Data Wrangler. Which approach should the scientist take?

Accepted Answer

Use AWS Glue DataBrew to create a profile job that outputs data quality reports. AWS Glue DataBrew provides a visual interface for data profiling and can handle large datasets without writing code. It automatically detects data types, missing values, and summary statistics. Option A is wrong because SageMaker Data Wrangler requires more manual setup and coding, and is not as straightforward for quick profiling. Option B is wrong because Amazon Athena requires writing SQL queries and is not a dedicated profiling tool. Option D is wrong because AWS Glue ETL with PySpark requires writing custom code.

Answer

Use Amazon SageMaker Data Wrangler to import the data and generate a report

Answer

Use Amazon Athena to run SELECT statements on each column

Answer

Use AWS Glue ETL jobs with PySpark to compute statistics

Question 17

A research lab is using SageMaker to train deep learning models on a custom dataset stored in S3. Each training job uses a single ml.p3.2xlarge instance. Recently, training jobs have been failing intermittently with 'NetworkError: Connection reset by peer' during the data download phase. The data scientist notices that the dataset is 50GB and the network throughput is low. The training script uses the default S3 download method (boto3) to copy data from S3 to the local instance storage. Which solution should the data scientist implement to resolve the issue?

Accepted Answer

Use SageMaker Pipe mode to stream data directly from S3.. SageMaker Pipe mode streams training data directly from S3 without writing to local disk, avoiding the large download that causes network timeouts. Option A (mount EBS) does not eliminate the need to download 50GB, so it does not resolve the network reset. Option C (retry logic) may help but does not address the root cause of low throughput and large downloads. Option D (larger instance) increases bandwidth but does not guarantee connection stability and costs more.

Answer

Mount an EBS volume to the instance and copy data there before training.

Answer

Add retry logic in the training script to handle network errors.

Answer

Use a larger instance type like p3.8xlarge for better network bandwidth.

Question 18

A data scientist is working on a regression problem with a dataset that contains outliers. The data scientist is choosing between mean squared error (MSE) and mean absolute error (MAE) as the loss function. Which loss function is more robust to outliers?

Accepted Answer

MAE, because it treats all errors equally.. MAE is more robust to outliers because it uses the absolute difference between predicted and actual values, which does not disproportionately penalize large errors. In contrast, MSE squares the errors, causing outliers to have a much larger influence on the loss and model updates. This makes MAE less sensitive to extreme values in regression tasks.

Answer

Both are equally robust to outliers.

Answer

MSE, because it penalizes large errors more heavily.

Answer

Neither is robust; use Huber loss instead.

Question 19

A company operates a real-time fraud detection system using SageMaker. The model is deployed on an ml.c5.xlarge instance behind an Application Load Balancer (ALB). Recently, during a sales event, traffic spiked and the endpoint returned HTTP 503 errors. The team scaled the instance count from 2 to 5, but errors persisted. CloudWatch metrics show low CPU utilization (~30%) and high memory usage (~90%). The model loads a large dictionary file (2GB) into memory at startup. Which action should resolve the issue?

Accepted Answer

Switch to a compute-optimized instance type like c5.2xlarge.. The high memory usage (~90%) with low CPU (~30%) indicates the instance is memory-constrained under load. The ml.c5.xlarge has 8 GB memory, and the model's 2 GB dictionary plus overhead exhausts memory, causing 503 errors. Scaling out doesn't help because each instance is individually memory-bound. Option B switches to c5.2xlarge, which provides 16 GB memory (doubling capacity) and more CPU cores, addressing both memory exhaustion and ensuring sufficient compute for concurrent requests. Option A (Spot instances) does not increase per-instance memory. Option C (scaling to 10 instances) still uses c5.xlarge instances with 8 GB each, so each remains memory-limited. Option D (r5.large) offers 16 GB memory but reduces vCPUs to 2, which could create a CPU bottleneck for the inference workload, making B the better choice.

Answer

Enable auto-scaling with Spot instances.

Answer

Increase the number of instances further to 10.

Answer

Use a memory-optimized instance type like r5.large.

Question 20

A data scientist is using Amazon SageMaker to train a model on a large dataset (10 TB) stored in S3 in Parquet format. The training job uses an ml.p3.16xlarge instance with multiple GPUs. The data scientist notices that the GPU utilization is low (around 30%) and the training is slow. The dataset consists of hundreds of thousands of small Parquet files. The data scientist suspects that the I/O is bottlenecked. What should the data scientist do to improve GPU utilization and training speed?

Accepted Answer

Consolidate the small Parquet files into larger files (e.g., 1 GB each). Consolidating small Parquet files into larger files (e.g., 1 GB each) reduces the overhead of reading many small files from S3, improving I/O throughput and keeping GPUs busy. Option A (increase batch size) may help GPU utilization but does not address the I/O bottleneck. Option C (use a smaller instance) would not improve speed and may worsen the situation. Option D (Pipe input mode) can help with streaming but does not solve the small file issue; the data still comes from many small files.

Answer

Increase the batch size

Answer

Use a smaller instance type to reduce cost

Answer

Use Pipe input mode to stream data directly