DA0-002 Data Analysis Practice Test 5 — 15 Questions

Question 1

A healthcare analytics team is building a predictive model to identify patients at high risk of readmission within 30 days of discharge. The dataset includes 50,000 patient records with 200 features, including demographics, vital signs, lab results, and historical admissions. The target variable is binary (readmitted or not). The team uses a logistic regression model and achieves an AUC of 0.72 on the test set. However, the model's calibration is poor: for patients predicted to have a 70% risk, the actual readmission rate is only 40%. The team wants to improve calibration without significantly reducing discrimination (AUC). The data scientist suggests applying Platt scaling. However, the team lead is concerned that Platt scaling may reduce the model's ability to rank patients correctly. Which of the following is the best course of action?

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

Accepted Answer

Apply Platt scaling on a held-out validation set to recalibrate the predicted probabilities without refitting the original model.. Platt scaling is a post-processing technique that fits a logistic regression model on the predicted probabilities from the original model using a held-out validation set. This recalibrates the probabilities without altering the ranking of patients (the AUC remains unchanged), directly addressing the poor calibration while preserving discrimination. Option C correctly describes this procedure.

Answer

Remove poorly calibrated predictions by discarding all patients with predicted risk between 0.3 and 0.7.

Answer

Ignore calibration because AUC is the only metric that matters for readmission risk models.

Answer

Switch to a random forest model, which inherently produces better-calibrated probabilities.

Question 2

A data analyst at a marketing firm is tasked with segmenting customers based on their purchasing behavior. The dataset contains 10,000 customers with features such as annual spend, frequency of purchases, recency of last purchase, and average order value. The analyst decides to use k-means clustering. After standardizing the features, the analyst runs k-means with k=3, k=4, and k=5, and computes the silhouette score for each: k=3: 0.45, k=4: 0.52, k=5: 0.48. The analyst also plots the elbow curve and observes that the within-cluster sum of squares (WCSS) decreases sharply from k=2 to k=4, then levels off. Based on these results, what is the most appropriate number of clusters?

Accepted Answer

k=4. The silhouette score is highest at k=4 (0.52), indicating that clusters are well-separated and cohesive. The elbow curve shows WCSS decreasing sharply up to k=4 and then leveling off, suggesting that k=4 captures the optimal trade-off between model complexity and variance explained. Together, these metrics point to k=4 as the most appropriate number of clusters.

Answer

k=2

Answer

k=3

Answer

k=5

Question 3

A healthcare analytics team is building a predictive model to identify patients at high risk of readmission within 30 days of discharge. The dataset includes 50,000 patient records with 200 features, including demographics, vital signs, lab results, and historical admissions. The target variable is binary (readmitted or not). The team uses a logistic regression model and achieves an AUC of 0.72 on the test set. However, the model's calibration is poor: for patients predicted to have a 70% risk, the actual readmission rate is only 40%. The team wants to improve calibration without significantly reducing discrimination (AUC). The data scientist suggests applying Platt scaling. However, the team lead is concerned that Platt scaling may reduce the model's ability to rank patients correctly. Which of the following is the best course of action?

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

Accepted Answer

Apply SMOTE to oversample the readmission class. SMOTE (Synthetic Minority Oversampling Technique) directly addresses the class imbalance by generating synthetic samples for the minority class (readmission). This forces the logistic regression model to learn decision boundaries that separate the two classes, rather than defaulting to the majority class prediction. With 85% majority and 15% minority, accuracy alone is misleading, and SMOTE is a proven technique to improve recall for the minority class.

Answer

Remove features with low variance to reduce noise

Answer

Use accuracy as the evaluation metric to monitor improvement

Answer

Switch to a random forest model with default settings

Question 4

Drag and drop the steps to normalize a database table from 1NF to 3NF in the correct order.

Accepted Answer

Step 1: Achieve 1NF by removing repeating groups and ensuring atomic values. Step 2: Achieve 2NF by removing partial dependencies. Step 3: Achieve 3NF by removing transitive dependencies.. Normalization proceeds from 1NF to 2NF to 3NF, then table creation and foreign keys.

Answer

Step 1: Achieve 3NF by removing transitive dependencies. Step 2: Achieve 2NF by removing partial dependencies. Step 3: Achieve 1NF by removing repeating groups.

Answer

Step 1: Achieve 2NF by removing partial dependencies. Step 2: Achieve 1NF by removing repeating groups. Step 3: Achieve 3NF by removing transitive dependencies.

Answer

Step 1: Achieve 1NF by removing repeating groups. Step 2: Achieve 3NF by removing transitive dependencies. Step 3: Achieve 2NF by removing partial dependencies.

Question 5

Drag and drop the steps to implement a data classification policy in the correct order.

Accepted Answer

Define classification levels, assign ownership, label data, enforce access control, train employees. Classification involves defining levels, assigning ownership, labeling, access control, and training.

Answer

Train employees, define classification levels, assign ownership, label data, enforce access control

Answer

Label data, define classification levels, assign ownership, enforce access control, train employees

Answer

Define classification levels, label data, assign ownership, enforce access control, train employees

Question 6

Match each data sampling method to its description.

Accepted Answer

Simple Random Sampling: Every member of the population has an equal chance of being selected.. In matching, ensure you distinguish between methods: simple random gives equal chance; stratified divides by characteristic; cluster divides into groups and picks whole groups; systematic uses a fixed interval.

Answer

Simple Random Sampling: Population is divided into clusters, and entire clusters are randomly selected.

Answer

Stratified Sampling: Every kth member is selected from a list after a random start.

Question 7

A data analyst is designing a data model for a sales data warehouse. The model should optimize query performance for aggregations by minimizing joins and duplicating data where necessary. Which schema design should the analyst use?

Accepted Answer

Star schema. A star schema is the correct choice because it organizes data into a central fact table surrounded by denormalized dimension tables, which minimizes the number of joins required for aggregation queries. By duplicating dimension attributes rather than normalizing them, the star schema trades storage for query speed, making it ideal for data warehouse workloads that emphasize analytical performance. This design directly supports the requirement to optimize aggregations while reducing join complexity.

Answer

Entity-relationship model

Answer

Snowflake schema

Answer

3NF normalized model

Question 8

A data scientist is building a predictive model to forecast monthly sales. The data shows a linear trend with no seasonality. Which regression technique is most appropriate?

Accepted Answer

Linear regression. Linear regression is the most appropriate technique because the data shows a linear trend with no seasonality, making a straight-line model the simplest and most effective fit. It directly models the relationship between the independent variable (e.g., time) and the dependent variable (monthly sales) using a linear equation, minimizing the sum of squared residuals.

Answer

Polynomial regression

Answer

Logistic regression

Answer

Ridge regression

Question 9

A data analyst is working with a dataset containing house prices. After building a multiple linear regression model, the analyst observes that the model performs well on training data but poorly on validation data. Which technique is most appropriate to address this issue?

Accepted Answer

Apply L2 regularization (Ridge). The model is overfitting the training data, as evidenced by high performance on training data but poor performance on validation data. L2 regularization (Ridge) adds a penalty term proportional to the square of the coefficients, which shrinks them and reduces model complexity, thereby improving generalization to unseen data.

Answer

Decrease the training data size

Answer

Use a polynomial transformation

Answer

Increase the number of features

Question 10

A marketing team wants to segment customers into distinct groups based on purchasing behavior. The data includes numeric features such as frequency, monetary value, and recency. Which unsupervised learning algorithm should be used?

Accepted Answer

K-means clustering. K-means clustering is the correct choice because it is an unsupervised learning algorithm that partitions data into K distinct clusters based on feature similarity. For segmenting customers by purchasing behavior (frequency, monetary value, recency), K-means groups customers with similar numeric patterns without requiring labeled outcomes, making it ideal for exploratory segmentation.

Answer

Decision tree

Answer

Linear regression

Answer

Association rules

Question 11

A data analyst is building a binary classification model to predict customer churn. The dataset is imbalanced, with only 10% churners. The analyst wants to evaluate model performance with a focus on correctly identifying churners. Which metric is most appropriate?

Accepted Answer

Recall (sensitivity). Recall (sensitivity) is the most appropriate metric because it measures the proportion of actual churners correctly identified by the model. Since the dataset is imbalanced (only 10% churners) and the analyst's focus is on correctly identifying churners, recall directly addresses the cost of missing positive cases (false negatives). Accuracy would be misleading due to class imbalance, while precision and F1-score prioritize different trade-offs.

Answer

F1-score

Answer

Precision

Answer

Accuracy

Question 12

A data analyst needs to join two tables in a SQL database: Orders and Customers. The analyst wants to include all orders, even if there is no matching customer record. Which type of join should be used?

Accepted Answer

LEFT JOIN. A LEFT JOIN returns all rows from the left table (Orders) and the matching rows from the right table (Customers). If there is no match, NULL values are returned for the right table's columns. This satisfies the requirement to include all orders, even those without a matching customer record.

Answer

RIGHT JOIN

Answer

FULL OUTER JOIN

Answer

INNER JOIN

Question 13

A data analyst is analyzing survey responses where respondents rated satisfaction on a scale of 1-5. The analyst wants to visualize the distribution of responses. Which chart type is most appropriate?

Accepted Answer

Histogram. A histogram is the most appropriate chart for visualizing the distribution of a single discrete variable, such as satisfaction ratings on a 1-5 scale. It groups the responses into bins (each rating value) and displays the frequency of each bin using bars, clearly showing the shape, central tendency, and spread of the data.

Answer

Box plot

Answer

Scatter plot

Answer

Line chart

Question 14

Which TWO of the following are dimensional modeling techniques commonly used in data warehouses?

Accepted Answer

Snowflake schema. Snowflake schema (B) is a dimensional modeling technique in which dimension tables are normalized into multiple related tables, reducing redundancy while preserving the fact-dimension structure of a data warehouse. Star schema (C) is the classic dimensional modeling technique where a central fact table joins directly to denormalized dimension tables, optimizing query performance for analytical workloads. Both are standard approaches described by Kimball for organizing data marts and warehouses around facts and dimensions. Entity-relationship diagram (A) is a conceptual/logical modeling notation for OLTP-style normalized databases, not a dimensional technique. Scatter plot (D) and histogram (E) are data visualization or statistical analysis tools, not data warehouse modeling techniques.

Answer

Entity-relationship diagram

Answer

Scatter plot

Answer

Histogram

Question 15

Which THREE of the following are common steps in data cleaning?

Accepted Answer

Imputing missing values. Imputing missing values is a common data cleaning step because real-world datasets often have gaps due to data collection errors or system failures. Techniques like mean/median imputation, regression imputation, or using algorithms like k-NN help preserve sample size and avoid bias that would result from simply dropping rows. This ensures the dataset remains usable for analysis without introducing significant distortion.

Answer

Removing outliers without justification

Answer

Increasing sample size