Hard — DA0-002 Practice Quiz

Question 1

A pharmaceutical analyst is preparing a dataset for a regulatory audit. The auditor asks the analyst to document not just how the data moved and transformed through the pipeline, but also the original source of the data, who collected it, the methods used, and any custody transfers before it entered the company's systems. Which data governance concept specifically addresses this requirement for documenting the origin and chain of custody of the data?

Accepted Answer

Data provenance

Answer

Data lineage

Answer

Data cataloging

Answer

Data stewardship

Question 2

A data analyst is preparing a dataset of monthly customer transaction amounts for a machine learning model. Profiling reveals a small number of legitimate but extremely large B2B purchases that inflate the mean and skew the distribution. The business confirms these transactions are valid and must remain represented in the analysis, but their extreme magnitude distorts the model's sensitivity to typical retail purchases. Which technique best addresses the outliers while satisfying the business requirement?

Accepted Answer

Apply winsorization to cap the extreme values at a defined percentile threshold

Answer

Delete the rows containing the extreme B2B transaction values

Answer

Replace the extreme values with the column mean

Answer

Convert the transaction amount column to a categorical variable

Question 3

A retail analyst runs a linear regression to test whether an increase in store parking spaces predicts daily foot traffic. The model returns a slope coefficient of 4.2 with a 95% confidence interval of [-1.8, 10.2] and a p-value of 0.14. Management asks whether they can conclude that adding parking spaces increases foot traffic. What is the most accurate interpretation?

Accepted Answer

The relationship is not statistically significant because the confidence interval includes zero and the p-value exceeds 0.05.

Answer

The relationship is significant because the slope coefficient is positive at 4.2.

Answer

The relationship is significant because the confidence interval is wide, indicating a strong effect.

Answer

The negative lower bound proves that parking spaces reduce foot traffic.

Question 4

A retail analyst runs a simple linear regression predicting monthly sales from advertising spend. The model reports a slope coefficient with a p-value of 0.002 and an R-squared of 0.18. The marketing director wants to know what these two results together tell them about the relationship. What is the most accurate interpretation?

Accepted Answer

Advertising spend has a statistically significant relationship with sales, but it explains only about 18% of the variation in sales.

Answer

The relationship is not statistically significant because R-squared is well below 0.50.

Answer

Advertising spend explains 82% of the variation in sales, and the low p-value confirms this strong fit.

Answer

Because the p-value is significant, the model can reliably predict individual monthly sales figures with high accuracy.

Question 5

A retail analyst builds a regression model to forecast next quarter's sales for a single new store based on square footage. Management asks for a range that captures where that specific store's actual sales are likely to fall, not just the average sales of all stores of that size. Which interval should the analyst report, and why?

Accepted Answer

A prediction interval, because it accounts for both the uncertainty in the estimated mean and the variability of individual observations around that mean

Answer

A confidence interval, because it always produces a wider range that better protects against error for a single store

Answer

A confidence interval, because it estimates the range for an individual outcome rather than the average

Answer

A prediction interval, because it narrows the range by ignoring the residual variability of individual data points

Question 6

An analyst builds a linear regression model predicting monthly energy consumption from average outdoor temperature. After fitting the model, they plot the residuals against the predicted values and notice the points form a clear U-shaped (curved) pattern rather than being randomly scattered around zero. What does this pattern most likely indicate about the model?

Accepted Answer

The relationship between the variables is nonlinear, violating the linearity assumption

Answer

The residuals are normally distributed, confirming the model is valid

Answer

There is strong multicollinearity between the predictor variables

Answer

The sample size is too small to produce reliable estimates

Question 7

A hospital analyst compares two treatment programs. Looking at the combined dataset, Program A has a higher overall recovery rate than Program B. However, when the analyst splits the data by patient severity (mild vs. severe cases), Program B actually has a higher recovery rate within BOTH the mild group and the severe group. What phenomenon is the analyst observing, and what should guide the conclusion?

Accepted Answer

Simpson's paradox; the segmented (subgroup) results are more valid because a lurking variable—case severity—is confounding the aggregated comparison

Answer

Sampling error; the analyst should increase the sample size until the aggregate and subgroup results agree

Answer

Multicollinearity; the analyst should drop the severity variable to resolve the contradiction

Answer

Regression to the mean; the aggregated result is correct and the subgroup differences will disappear over time

Question 8

A data architect is designing a data warehouse for a retail analytics team. The product dimension currently duplicates category, subcategory, and supplier names across millions of rows, and storage costs are rising. The team is willing to accept slightly more complex queries to reduce this redundancy. Which schema design should the architect adopt?

Accepted Answer

Snowflake schema, normalizing dimension tables into related sub-tables

Answer

Star schema, keeping all dimension attributes denormalized in single tables

Answer

Flat file schema, storing all facts and dimensions in one wide table

Answer

OLTP third-normal-form schema optimized for transaction inserts