MLS-C01 Exploratory Data Analysis Practice Test 6 — 15 Questions

Question 1

A data scientist is exploring a dataset with 10 features and observes that the correlation between feature A and feature B is 0.98. Which action should be taken to address multicollinearity before training a linear regression model?

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

Accepted Answer

Remove one of the two features from the dataset.. Dropping one of the highly correlated features reduces redundancy and mitigates multicollinearity. Option A (PCA) creates orthogonal components, which addresses multicollinearity but reduces interpretability; dropping a feature is more straightforward. Option B (Min-Max scaling) does not address collinearity at all. Option D is wrong because adding polynomial features increases correlation.

Answer

Use Principal Component Analysis (PCA) to combine them.

Answer

Apply Min-Max scaling to both features.

Answer

Add polynomial features to both.

Question 2

A machine learning engineer trains a binary classifier on an imbalanced dataset where the positive class represents 1% of the data. After training, the model achieves 99% accuracy but only 10% recall on the positive class. Which metric should the engineer focus on to evaluate the model's performance on the minority class?

Accepted Answer

F1 score. (F1 score) is the correct metric because it balances precision and recall, providing a single measure that is robust to class imbalance. With only 1% positive class, accuracy (Option B) is misleadingly high due to the majority class. AUC-ROC (Option C) can still be high even if recall is low, as it evaluates ranking rather than absolute performance. Precision (Option D) only considers the proportion of correct positive predictions, ignoring false negatives, which is not suitable when recall is poor. The F1 score captures both aspects, making it the best choice for evaluating minority class performance in this scenario.

Answer

Accuracy

Answer

AUC-ROC

Answer

Precision

Question 3

A data engineer runs a SQL query on Amazon Athena to explore a dataset stored in S3 as CSV. The query returns zero rows for a column that should have numeric values. Which step should the engineer take to diagnose the issue?

Accepted Answer

Check the table schema in AWS Glue Data Catalog to ensure the column data type is correct.. Checking the table schema in the AWS Glue Data Catalog helps identify data type mismatches. If Athena returns zero rows for a numeric column, it may be because the column's data type in the catalog is incorrect (e.g., string instead of int). Fixing the schema to match the actual data allows Athena to parse the values correctly. Option A is incorrect because encryption does not affect query results. Option B is incorrect because running a crawler will only re-infer the schema, which may not solve the issue if the underlying data format is inconsistent. Option C is incorrect because partitioning is unrelated to data type issues.

Answer

Verify that the S3 bucket has encryption enabled.

Answer

Run an AWS Glue crawler to update the table schema.

Answer

Add a partition to the table for the date column.

Question 4

A data scientist is performing EDA on a dataset with missing values in 3 of 20 features. The missing rate is 5% for each feature. The scientist wants to preserve as much data as possible while avoiding bias. Which imputation strategy is most appropriate?

Accepted Answer

Impute missing values with the median of each feature.. Median imputation is the most appropriate choice because it preserves all rows (no data loss) while being robust to outliers and skewed distributions, which are common in real-world features. With only 5% missingness per feature, imputing the median introduces minimal bias and keeps the dataset intact for downstream modeling. It is also simple, deterministic, and does not assume a particular distribution, unlike mean imputation.

Answer

Remove rows with any missing values.

Answer

Impute missing values with the mean of each feature.

Answer

Use K-Nearest Neighbors (KNN) imputation.

Question 5

During EDA, a data scientist finds that a numeric feature has many outliers. The feature will be used in a linear regression model. Which approach should the scientist take to handle the outliers?

Accepted Answer

Cap the feature values at the 1st and 99th percentiles.. Capping (winsorizing) values at the 1st and 99th percentiles limits the influence of extreme outliers on the linear regression coefficients while retaining all rows and preserving the feature's overall distribution shape. Linear regression is sensitive to outliers because OLS minimizes squared error, so extreme values disproportionately pull the fitted line. Capping is a robust, standard EDA remediation that balances outlier control with data retention.

Answer

Remove all rows with outlier values.

Answer

Apply a logarithmic transformation to the feature.

Answer

Standardize the feature using Z-score normalization.

Question 6

A data scientist is exploring a dataset with 10 features and observes that the correlation between feature A and feature B is 0.98. Which action should be taken to address multicollinearity before training a linear regression model?

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

Accepted Answer

Use Amazon SageMaker Data Wrangler to import the dataset from S3 and generate a correlation matrix.. Amazon SageMaker Data Wrangler can directly import the dataset from S3 and generate a correlation matrix efficiently without needing to write custom code, making it the most efficient approach for identifying feature correlations with the target variable. Option B is incorrect because using Amazon QuickSight to create scatter plots for each of the 500 features would be time-consuming and not scalable. Option C is incorrect because Amazon Athena uses SQL queries which are not designed to compute correlation coefficients efficiently across a large number of features. Option D is incorrect because AWS Glue ETL is intended for data transformation pipelines and is not suitable for quick interactive correlation analysis.

Answer

Use Amazon QuickSight to create scatter plots for each feature vs. target.

Answer

Use Amazon Athena with SQL queries to compute correlation coefficients.

Answer

Use AWS Glue ETL to compute pairwise correlations and output to Amazon Redshift.

Question 7

A company is preparing a dataset for training a binary classification model. The dataset has a severe class imbalance (1% positive class). The data scientist wants to understand the impact of this imbalance on model performance before sampling. Which exploratory analysis step is MOST critical?

Accepted Answer

Plot the distribution of each feature separately for the positive and negative classes.. The most critical step because plotting the distribution of each feature separately for the positive and negative classes allows the data scientist to visually assess class separability, overlap, and feature behavior under severe imbalance. This insight directly informs the impact of imbalance on model performance before any sampling. Options A, B, and C are less critical at this stage: correlation with the target (A) does not reveal class-level distributions; missing values and outliers (B) are important but not specific to understanding imbalance impact; PCA (C) is a dimensionality reduction technique that may obscure per-feature patterns and is not necessary for initial exploratory analysis of class distributions.

Answer

Compute the correlation matrix of all features with the target variable.

Answer

Check for missing values and outliers in the dataset.

Answer

Perform PCA and visualize the first two principal components colored by class.

Question 8

A data scientist is exploring a dataset with 10 features and observes that the correlation between feature A and feature B is 0.98. Which action should be taken to address multicollinearity before training a linear regression model?

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

Accepted Answer

Yeo-Johnson transformation. The Yeo-Johnson transformation is a power transformation that can handle both positive and negative values, making it suitable for making skewed distributions more Gaussian-like. It is a generalization of the Box-Cox transformation, which only works for positive data. Since the dataset may contain negative values, Yeo-Johnson is the most appropriate choice.

Answer

Log transformation

Answer

Standard scaling (z-score normalization)

Answer

Box-Cox transformation

Question 9

An IAM policy is attached to a data scientist's role. The scientist is trying to list objects in the 'data-bucket' using Amazon Athena. The query fails with an access denied error. What is the MOST likely reason?

Accepted Answer

The query is trying to read data from the 'sensitive/' prefix.. The IAM policy likely includes a Deny statement for the 'sensitive/' prefix, causing access denied when Athena attempts to read data from that location. Option A is incorrect because s3:ListBucket is typically required and allowed; the error is access denied rather than a forbidden error. Option B is incorrect because a syntax error would usually produce an invalid policy error, not an access denied error during query execution. Option D is incorrect because if s3:GetObject were explicitly denied for all objects, the error would occur for all queries, but the issue is specific to the 'sensitive/' prefix.

Answer

The policy does not allow s3:ListBucket on the bucket.

Answer

The policy has a syntax error.

Answer

The s3:GetObject action is explicitly denied for all objects.

Question 10

A data engineer ingests streaming data into Amazon Kinesis Data Streams. The data science team needs to analyze the data using Amazon SageMaker notebooks. What is the most efficient way to provide access to the stream data for ad-hoc exploration?

Accepted Answer

Use the Kinesis connector for Spark to read data directly from the stream into a Spark DataFrame in the notebook.. Using the Kinesis connector for Spark allows the SageMaker notebook to read data directly from the Kinesis stream into a Spark DataFrame. This is the most efficient way for ad-hoc exploration because it provides real-time access to the stream without additional storage or transformation steps. It leverages the native integration between Spark and Kinesis.

Answer

Create an AWS Lambda function to transform and write data to DynamoDB, then query DynamoDB from the notebook.

Answer

Configure a Kinesis Firehose delivery stream to deliver data to an S3 bucket, then query the data from the notebook using Athena.

Answer

Install the Kinesis Agent on the SageMaker notebook instance and configure it to write data to a local file.

Question 11

A data analyst is performing EDA on a tabular dataset with 500 features. The goal is to reduce dimensionality before modeling. Which TWO techniques are appropriate for this task?

Accepted Answer

t-distributed Stochastic Neighbor Embedding (t-SNE).. PCA (Option B) and t-SNE (Option A) are both dimensionality reduction techniques. PCA is linear, while t-SNE is nonlinear. Option C (k-fold cross-validation) is used for model evaluation, not dimensionality reduction. Option D (Chi-square test) is a feature selection method for categorical targets, not a dimensionality reduction technique. Option E (one-hot encoding) increases the number of features, which is the opposite of dimensionality reduction.

Answer

k-fold cross-validation to assess model performance.

Answer

Chi-square test for independence between features.

Answer

One-hot encoding of categorical variables.

Question 12

Refer to the exhibit. A data scientist is unable to query a table in Amazon Athena that is located in the 'my-data-bucket' S3 bucket. The IAM policy shown is attached to the scientist's role. What is the most likely reason for the failure?

Accepted Answer

The policy does not allow s3:PutObject to write query results to an S3 bucket.. Amazon Athena requires an S3 bucket to store query results. The IAM policy must allow `s3:PutObject` on the results bucket. Without this permission, Athena cannot write query output, causing the query to fail. The policy shown likely lacks this permission, making D the most likely reason.

Answer

The policy does not allow decrypting data encrypted with AWS KMS.

Answer

The policy does not allow athena:StartQueryExecution.

Answer

The policy does not allow s3:GetObject on the bucket.

Question 13

Refer to the exhibit. A data scientist is using AWS Glue ETL jobs to process data from a source database. The job logs show repeated timeout errors. Which EDA step should the scientist perform to diagnose the issue?

Accepted Answer

Check the source database table sizes and row counts over time.. Repeated timeout errors in an AWS Glue ETL job reading from a source database most often stem from the volume of data being read — large tables or growing row counts exceed the job's read window or connection timeout. An exploratory data analysis (EDA) step of checking source table sizes and row counts over time reveals whether data growth is the cause and informs partitioning, filtering, or job-sizing changes. This is a diagnostic EDA action rather than a blind fix.

Answer

Test network connectivity from the Glue job to the source database using telnet.

Answer

Switch the Glue ETL job type from Spark to Python shell to reduce overhead.

Answer

Increase the Glue job timeout to 600 seconds and rerun.

Question 14

A data scientist is exploring a dataset and wants to identify outliers in a numerical feature. The feature is not normally distributed. Which technique is robust to non-normal distributions?

Accepted Answer

Use the IQR method: flag values below Q1 - 1.5*IQR or above Q3 + 1.5*IQR.. The IQR method is non-parametric — it relies on quartiles (Q1, Q3) rather than mean and standard deviation, so it does not assume normality. Flagging values below Q1 − 1.5×IQR or above Q3 + 1.5×IQR is the standard Tukey fence approach and remains robust even when the feature is skewed or heavy-tailed.

Answer

Compute the Median Absolute Deviation (MAD) and flag values with MAD > 3.

Answer

Calculate the Z-score and flag values with |Z| > 3.

Answer

Flag values more than 3 standard deviations from the mean.

Question 15

A data analyst is exploring a dataset with a target variable that is highly imbalanced. The minority class represents only 1% of the data. Which technique should the analyst use to better understand the relationships between features and the minority class?

Accepted Answer

Use stratified sampling to create a balanced sample for analysis.. Stratified sampling ensures the minority class is proportionally represented in the sample, allowing meaningful analysis. Option A is wrong because SMOTE generates synthetic data, which is not appropriate for initial exploratory analysis. Option B is wrong because random sampling may miss the minority class entirely. Option C is wrong because scaling features does not address class imbalance.

Answer

Apply SMOTE to the dataset before analysis.

Answer

Use random sampling to reduce the dataset size.

Answer

Scale the features using Min-Max scaling.