MLS-C01 Modeling Practice Test 20 — 15 Questions

Question 1

A data scientist is training a model using Amazon SageMaker's built-in k-means algorithm on a dataset with 10 million records and 100 features. The training job is taking a very long time. The scientist wants to reduce training time without significantly sacrificing model quality. Which action should the scientist take?

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

Accepted Answer

Use the 'mini_batch_size' hyperparameter to increase the batch size.. Increasing the mini_batch_size hyperparameter in Amazon SageMaker's k-means can improve computational efficiency by processing more examples per batch, thus reducing the number of updates and leveraging optimized linear algebra routines. This can lead to faster training without drastically altering the clustering outcome, provided the batch size is tuned appropriately.

Answer

Enable the 'verbose' hyperparameter to log more details.

Answer

Reduce the number of clusters (k) from 100 to 10.

Answer

Reduce the number of features by applying PCA before training.

Question 2

A fraud detection team trains a binary classifier in Amazon SageMaker on data where fraudulent transactions are only 0.5% of records. After training, the model shows very high accuracy but poor recall on the fraud class, and the team wants to improve detection of the minority class without collecting new data. Which two actions should they take? (Choose two.)

Accepted Answer

Use the SageMaker built-in XGBoost scale_pos_weight hyperparameter to give the positive class more weight.. Class imbalance makes the loss dominated by majority-class examples, so the model minimizes error by favoring the majority. Weighting the positive class through scale_pos_weight and generating synthetic minority examples with SMOTE both rebalance the effective training signal and typically raise recall. Maximizing training accuracy, discarding most majority data, or optimizing accuracy as the metric all fail to correct the underlying imbalance.

Answer

Increase the number of training epochs until training accuracy reaches 100%.

Answer

Switch the objective metric to accuracy and retrain the model.

Answer

Remove the majority-class records until the classes are equal in size.

Question 3

A machine learning engineer is using Amazon SageMaker to train a deep learning model for a natural language processing task. The model is trained on a large corpus and the engineer wants to reduce the vocabulary size by mapping rare words to a special token. Which SageMaker built-in algorithm or feature should the engineer use to achieve this?

Accepted Answer

Use the SageMaker BlazingText algorithm with the 'subwords' option enabled.. The BlazingText algorithm in Amazon SageMaker supports the 'subwords' option, which allows the model to handle rare words by representing them as subword units. This is an effective way to reduce the vocabulary size and improve generalization for rare words. The other algorithms do not offer this specific feature; they serve different purposes such as sequence-to-sequence modeling, embedding learning, or topic modeling.

Answer

Use the SageMaker Object2Vec algorithm with negative sampling.

Answer

Use the SageMaker Sequence-to-Sequence algorithm with a fixed vocabulary.

Answer

Use the SageMaker Latent Dirichlet Allocation (LDA) algorithm with a low topic count.

Question 4

A data scientist is training a linear regression model using Amazon SageMaker's built-in Linear Learner algorithm. The dataset contains a categorical feature with high cardinality (over 10,000 unique values). The data scientist wants to avoid the curse of dimensionality and reduce training time. Which preprocessing step should be applied to the categorical feature before training?

Accepted Answer

Apply target encoding (mean encoding).. Target encoding is the best choice for high-cardinality categorical features in linear models because it reduces dimensionality while preserving the predictive information. One-hot encoding would explode the feature space, and hashing can cause collisions that degrade model performance. The Linear Learner algorithm does not automatically handle high-cardinality features effectively, so explicit target encoding is needed.

Answer

Use the built-in Linear Learner's automatic categorical encoding.

Answer

One-hot encode the categorical feature.

Answer

Hash the categorical feature into a fixed number of buckets.

Question 5

A machine learning team is using Amazon SageMaker to train a binary classification model on a dataset with 100,000 records and 30 features. The team wants to improve model performance by using the built-in XGBoost algorithm. Which two hyperparameters should the team tune to reduce overfitting? (Choose two.)

Accepted Answer

lambda. In XGBoost, max_depth and lambda are direct regularization hyperparameters that control model complexity. Lowering max_depth reduces tree depth, preventing the model from fitting noise. Increasing lambda adds L2 regularization, penalizing large weights. Together, they effectively reduce overfitting. Other options like learning_rate and subsample also affect overfitting but are less direct.

Answer

learning_rate

Answer

num_round

Answer

subsample

Question 6

A financial institution is building a fraud detection model using Amazon SageMaker. The dataset is highly imbalanced, with only 0.5% of transactions being fraudulent. The data scientist trains an XGBoost model using the built-in algorithm and evaluates it using accuracy, which is 99.5%. However, the model fails to identify any fraudulent transactions in the test set. The data scientist needs to improve the model's ability to detect fraud. Which approach should be taken?

Accepted Answer

Use the scale_pos_weight hyperparameter to account for class imbalance.. The model's high accuracy is misleading due to class imbalance. The scale_pos_weight hyperparameter in XGBoost is designed to handle imbalanced datasets by weighting the positive class more heavily. This encourages the model to learn patterns associated with fraud, improving recall. Other options either do not address the imbalance or are less practical within SageMaker's built-in algorithm.

Answer

Use a different evaluation metric such as AUC-ROC instead of accuracy.

Answer

Apply SMOTE to oversample the minority class before training.

Answer

Increase the number of boosting rounds to improve accuracy.

Question 7

A machine learning engineer is training a deep neural network for image classification using Amazon SageMaker. The training dataset contains 1 million images, and the engineer wants to use data augmentation to improve model generalization. The engineer decides to use the SageMaker TensorFlow estimator with a custom training script. The script currently reads the entire dataset into memory, but the training job fails with an out-of-memory error. The engineer needs to modify the data loading approach to handle the large dataset efficiently. Which solution is MOST appropriate?

Accepted Answer

Use the tf.data.Dataset API to create a pipeline that reads data from Amazon S3 in batches and applies augmentation on the fly.. The tf.data.Dataset API is designed for building scalable input pipelines that load and preprocess data in batches, avoiding the need to hold the entire dataset in memory. It supports parallel reads from S3 and on-the-fly augmentation, making it ideal for large image datasets. Other options either do not address the memory issue directly, are not compatible with custom TensorFlow scripts, or introduce inflexible preprocessing. The tf.data pipeline is the most efficient and standard solution.

Answer

Increase the instance type of the training job to one with more memory, such as an ml.p3.8xlarge, to accommodate the full dataset in memory.

Answer

Use SageMaker's Pipe mode input to stream data directly from S3 to the training container, bypassing the need to download the full dataset.

Answer

Preprocess the dataset by resizing all images to a smaller resolution and store the augmented versions in Amazon S3 before training.

Question 8

A machine learning engineer is using Amazon SageMaker to train a model with the built-in Object2Vec algorithm to generate embeddings for a product recommendation system. The engineer wants to ensure that the embeddings capture meaningful semantic relationships. The training data consists of pairs of products that are frequently bought together. Which two hyperparameters should the engineer tune to improve the quality of the embeddings? (Choose two.)

Accepted Answer

similarity_type. The quality of embeddings in Object2Vec depends on the embedding dimension and the similarity function used. enc_dim determines the capacity of the embedding space, while similarity_type defines how relationships are measured. Tuning these two hyperparameters directly impacts the model's ability to capture meaningful semantic relationships between products.

Answer

mini_batch_size

Answer

epochs

Answer

learning_rate

Question 9

A company is using Amazon SageMaker to train a deep learning model for image classification. The training job is taking too long, and the data scientist wants to speed up training without sacrificing model accuracy. The training uses a large dataset stored in Amazon S3 and a single GPU instance. Which two actions should the data scientist take to reduce training time? (Choose two.)

Accepted Answer

Use SageMaker distributed training with data parallelism to split the dataset across multiple GPUs and synchronize gradients.. To reduce training time for a large dataset on a single GPU, addressing I/O and compute bottlenecks is key. Streaming data with Pipe mode avoids the initial download delay and keeps the GPU fed. Distributed data parallelism leverages multiple GPUs to process more data in parallel, reducing the number of iterations per epoch. Both actions directly accelerate training without inherently compromising accuracy, provided hyperparameters like learning rate are adjusted appropriately for distributed training.

Answer

Increase the batch size to the maximum allowed by the GPU memory to reduce the number of iterations per epoch.

Answer

Use SageMaker Automatic Model Tuning to find the optimal hyperparameters that minimize training time.

Answer

Convert the dataset to the RecordIO format and use SageMaker's Pipe mode to improve I/O efficiency.

Question 10

A team is building a product recommendation system using matrix factorization in Amazon SageMaker. They notice that the model's training loss decreases steadily but validation loss starts increasing after 5 epochs. What is the most likely cause?

Accepted Answer

Overfitting. In matrix factorization for recommendation systems, a decreasing training loss with an increasing validation loss after several epochs is a classic sign of overfitting. The model is memorizing the training data (including noise) rather than learning generalizable patterns, which degrades its performance on unseen validation data.

Answer

Underfitting

Answer

Not enough training data

Answer

Learning rate too high

Question 11

A company is using Amazon SageMaker to train a deep learning model on a large dataset. The training job is taking too long. The team wants to reduce training time without changing the model architecture. Which action should they take?

Accepted Answer

Use SageMaker's distributed training with multiple instances. SageMaker's distributed training with multiple instances splits the dataset and model computations across several machines, enabling parallel processing that significantly reduces wall-clock training time. This approach leverages data parallelism or model parallelism without altering the model architecture, directly addressing the need for faster training.

Answer

Increase the learning rate by a factor of 10

Answer

Reduce the number of epochs

Answer

Reduce the batch size

Question 12

A data scientist is deploying a regression model in Amazon SageMaker that predicts housing prices. The model shows high bias (underfitting). Which action is most likely to reduce bias?

Accepted Answer

Add more features or increase model complexity. High bias (underfitting) means the model is too simple to capture the underlying patterns in the data. Adding more features or increasing model complexity (e.g., using polynomial features, deeper trees, or a more flexible algorithm) directly addresses underfitting by giving the model greater capacity to learn from the data. In Amazon SageMaker, this could involve using a more complex built-in algorithm like XGBoost with deeper trees or adding feature engineering transformations in a processing job.

Answer

Reduce the amount of training data

Answer

Increase regularization strength

Answer

Use a simpler model

Question 13

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 14

An e-commerce company uses a linear regression model to predict customer lifetime value (LTV). The model shows high variance on the test set, with training RMSE much lower than test RMSE. Which of the following is the MOST effective approach to reduce overfitting?

Accepted Answer

Apply L2 regularization (Ridge regression). High variance (low training RMSE, high test RMSE) indicates overfitting. L2 regularization (Ridge regression) adds a penalty proportional to the square of the coefficients, shrinking them toward zero without eliminating them, which reduces model complexity and improves generalization. This directly addresses overfitting by constraining the model's sensitivity to noise in the training data.

Answer

Use a polynomial kernel in a support vector regressor

Answer

Add more features, including interaction terms

Answer

Increase training data size by duplicating existing samples

Question 15

A company wants to use Amazon SageMaker to train a deep learning model using a custom TensorFlow script. The data is stored in an S3 bucket. Which SageMaker API operation should be used to launch the training job?

Accepted Answer

CreateTrainingJob. The correct API operation to launch a training job in Amazon SageMaker is CreateTrainingJob. This operation specifies the training algorithm (or custom script), resource configuration (instance type and count), input data configuration (pointing to S3), and output location for the model artifacts. It directly initiates the training process on SageMaker-managed infrastructure.

Answer

CreateHyperParameterTuningJob

Answer

CreateEndpoint

Answer

CreateTransformJob