DA0-002 Practice Exam 2 — 20 Questions

Question 1

In Looker Studio, what is the difference between dimensions and metrics?

Accepted Answer

Dimensions are categorical; metrics are numerical. In Looker Studio, dimensions are fields that contain categorical data (e.g., text, dates, or geographic names) used to group and segment data, while metrics are numerical fields (e.g., counts, sums, averages) that can be aggregated. Option D is correct because this distinction is fundamental to how Looker Studio processes and visualizes data: dimensions define the rows or categories in a chart, and metrics provide the quantitative values to be measured.

Answer

Dimensions are used for aggregation; metrics are for grouping

Answer

Dimensions are numerical; metrics are categorical

Answer

Both can be categorical or numerical

Question 2

A data analyst at a marketing firm is creating a weekly performance report for the marketing team. The report includes metrics like click-through rates, conversion rates, and cost per acquisition. The team prefers a quick overview of the week's performance to identify trends and make decisions in their Monday morning meeting. The meeting is only 30 minutes, and the team has limited time to review data. The analyst wants to provide a report that is concise and actionable. What format should the analyst use?

Accepted Answer

A one-page executive summary with key metrics highlighted.. A one-page executive summary with key metrics highlighted provides the marketing team with a concise, actionable overview that can be quickly reviewed in a 30-minute meeting. This format aligns with the requirement for a quick overview to identify trends and make decisions without overwhelming the team with excessive detail or requiring interactive exploration.

Answer

A live dashboard with interactive filters.

Answer

A detailed spreadsheet with all data.

Answer

A 20-slide presentation.

Question 3

A data governance team is drafting a policy for handling personally identifiable information (PII). According to data governance best practices, which document should define the classification levels and handling procedures?

Accepted Answer

Data classification policy. The data classification policy is the authoritative document that defines classification levels (e.g., public, internal, confidential, restricted) and specifies handling procedures for each category, including PII. This aligns with data governance best practices, as it establishes the rules for labeling, storing, transmitting, and disposing of sensitive data. A data dictionary describes metadata and schema, not classification rules.

Answer

Data dictionary

Answer

Data quality report

Answer

Data flow diagram

Question 4

A data analyst is profiling a dataset and finds that the 'email' column contains some NULL values. Which SQL query can be used to count how many rows have a NULL email?

Accepted Answer

SELECT COUNT(*) FROM table WHERE email IS NULL. SELECT COUNT(*) FROM table WHERE email IS NULL correctly counts rows where email is NULL. COUNT(*) counts all rows in the filtered result set, and IS NULL is the only valid comparison operator for NULL in SQL. This is the standard, portable way to count NULLs.

Answer

SELECT COUNT(email) FROM table WHERE email = NULL

Answer

SELECT SUM(CASE WHEN email IS NULL THEN 1 END) FROM table

Answer

SELECT COUNT(ISNULL(email)) FROM table

Question 5

Which TWO of the following are best practices for designing an accessible data visualization? (Choose 2.)

Accepted Answer

Add text labels or patterns to differentiate elements. Option A is correct because adding text labels or patterns (e.g., hatching, shapes, or direct data labels) provides a non-color-dependent way to distinguish data series, which is essential for users with color vision deficiencies or when charts are printed in grayscale. Option E is correct because using colorblind-friendly palettes (e.g., Okabe-Ito or ColorBrewer's colorblind-safe schemes) ensures that the chosen colors remain distinguishable for people with deuteranopia, protanopia, or tritanopia, satisfying WCAG 1.4.1 (Use of Color). Option B is incorrect because relying solely on color to convey information fails accessibility guidelines and excludes users who cannot perceive certain color differences. Option C is incorrect because 3D effects distort data perception, reduce readability, and add visual clutter without improving accessibility. Option D is incorrect because animated transitions can trigger motion sensitivity issues and are not a recognized accessibility best practice for data visualization.

Answer

Rely solely on color to convey information

Answer

Use 3D effects to make charts visually appealing

Answer

Include animated transitions between views

Question 6

A data analyst is preparing a presentation for executive leadership. The analyst wants to highlight the correlation between marketing spend and revenue over the past year. Which visualization type is most appropriate for showing this relationship?

Accepted Answer

Scatter plot. A scatter plot is the most appropriate visualization for showing the relationship between two continuous variables—marketing spend and revenue—because it plots individual data points on an X-Y axis, allowing the analyst to visually assess correlation, trends, and outliers. This directly supports the goal of highlighting correlation, as the pattern of points (e.g., upward slope) indicates the strength and direction of the relationship.

Answer

Pie chart

Answer

Bar chart

Answer

Histogram

Question 7

Drag and drop the steps to clean a dataset with missing values in the correct order.

Accepted Answer

Step 1: Identify missing values, Step 2: Analyze missingness pattern, Step 3: Select imputation/removal strategy, Step 4: Apply the strategy, Step 5: Document changes. Cleaning starts with identification, then analysis of missingness, strategy selection, application, and documentation.

Answer

Step 1: Analyze missingness pattern, Step 2: Identify missing values, Step 3: Select imputation/removal strategy, Step 4: Apply the strategy, Step 5: Document changes

Answer

Step 1: Identify missing values, Step 2: Select imputation/removal strategy, Step 3: Analyze missingness pattern, Step 4: Apply the strategy, Step 5: Document changes

Answer

Step 1: Identify missing values, Step 2: Analyze missingness pattern, Step 3: Apply the strategy, Step 4: Document changes, Step 5: Select imputation/removal strategy

Question 8

In pandas, you have a DataFrame 'df' with columns 'product' and 'sales'. You want to calculate the total sales per product. Which method should you use?

Accepted Answer

df.groupby('product')['sales'].sum(). The groupby('product')['sales'].sum() pattern is the canonical pandas idiom for split-apply-combine aggregation: it groups rows by the 'product' column, selects the 'sales' Series, and applies the sum aggregator to each group, returning a Series indexed by product. This is the most direct and efficient way to compute total sales per product.

Answer

df['sales'].apply(sum)

Answer

df.pivot_table(values='sales', index='product', aggfunc='sum')

Answer

df.merge(df, on='product')

Question 9

A data analyst at a subscription media company is asked to determine whether a new recommendation engine increased average watch time. The analyst plans a two-sample t-test comparing a randomly assigned control group against a randomly assigned treatment group, with watch time recorded in minutes per user per week. Which TWO conditions must hold for the pooled-variance two-sample t-test to be statistically valid here? (Choose two.)

Accepted Answer

The population variances must be equal, or the sample sizes must be large enough for the test's robustness.. The pooled two-sample t-test rests on independent observations and an assumption about variance equality that is relaxed by large, balanced samples. Random assignment to control and treatment satisfies independence, and checking variance similarity or group balance safeguards the test's nominal error rate. Sample-size thresholds and measurement-scale restrictions are not assumptions of this procedure, so they do not belong in the analyst's validation checklist.

Answer

Both groups must contain at least 30,000 users to satisfy the central limit theorem.

Answer

Watch time must be measured on a ratio scale with a true zero and no upper bound.

Answer

The control and treatment groups must have identical means before the experiment begins.

Question 10

A data analyst needs to visualize sales per capita across U.S. states. States with small populations but high sales (e.g., Delaware) appear too prominent on a choropleth map. Which technique best addresses this issue?

Accepted Answer

Use a choropleth map with rates instead of raw sales. Using a choropleth map with rates (e.g., sales per capita) instead of raw sales normalizes the data by population, preventing states with small populations from appearing overly prominent. This addresses the issue where Delaware, with high sales but low population, dominates the map. Rates allow for fair comparison across states regardless of population size.

Answer

Switch to a bar chart sorted by sales

Answer

Use a bubble chart with size proportional to sales

Answer

Apply a log scale to the color gradient

Question 11

A marketing analyst wants to segment customers based on their purchase history, including total spent, number of transactions, and average order value. The analyst runs k-means clustering with k=5 on the raw data but notices that the cluster assignments change significantly every time the algorithm is executed. What should the analyst do first to obtain consistent and meaningful clusters?

Accepted Answer

Normalize the features and set a fixed random seed for the initial centroids.. The instability in cluster assignments is caused by the algorithm's sensitivity to the scale of features and the random initialization of centroids. Normalizing the features ensures that each variable contributes equally to the distance calculations, while setting a fixed random seed makes the initial centroid selection deterministic, leading to reproducible results.

Answer

Switch to hierarchical clustering, which does not require specifying k.

Answer

Increase the number of clusters to k=10 to capture more detail.

Answer

Use principal component analysis (PCA) to reduce the number of features to two.

Question 12

A data analyst needs to present findings about customer churn to business stakeholders. The analysis identified that churn is highest among customers who have called customer support more than three times in the last month. Which of the following is the best way to communicate this insight?

Accepted Answer

A bar chart comparing churn rates for different support call counts.. A bar chart directly compares churn rates across discrete categories of support call counts (e.g., 0, 1, 2, 3, 4+ calls), making it easy for stakeholders to see the spike at 'more than three calls'. This aligns with the insight that churn is highest among customers with >3 support calls, and a bar chart is the standard visualization for comparing a continuous metric (churn rate) across categorical bins.

Answer

A scatter plot to show the relationship between support calls and churn.

Answer

A pie chart showing the proportion of churned vs. retained customers.

Answer

A table of raw churn data by customer ID.

Question 13

In time series decomposition, a pattern that repeats at regular intervals (e.g., weekly, yearly) is called:

Accepted Answer

Seasonality. Seasonality refers to a pattern that repeats at fixed, calendar-based intervals such as weekly, monthly, quarterly, or yearly. Because the question specifies 'regular intervals' like weekly or yearly, this is the defining characteristic of the seasonal component in time series decomposition. Seasonality is distinct from trend (long-term direction) and cyclical (irregular multi-year fluctuations).

Answer

Cyclical

Answer

Irregular

Answer

Trend

Question 14

A data scientist trains a regression model and observes high variance with low bias. Which technique is most appropriate to reduce variance?

Accepted Answer

Apply Ridge regularization. Ridge regularization (L2) reduces variance by adding a penalty term proportional to the square of the coefficients, which shrinks them toward zero without eliminating them. This directly addresses high variance (overfitting) by constraining the model's complexity, while low bias indicates the model fits the training data well. The regularization parameter λ controls the trade-off between bias and variance.

Answer

Increase polynomial features

Answer

Use a smaller training set

Answer

Remove correlated features

Question 15

Which THREE of the following are properties of ratio data? (Choose THREE.)

Accepted Answer

Supports multiplication and division. Ratio data supports multiplication and division because it has a true, meaningful zero point that indicates the absence of the measured attribute. This allows ratios to be computed (e.g., one value is twice another), which is a defining property of ratio scales in measurement theory.

Answer

Data can be categorized into groups

Answer

Allows negative values

Question 16

An analyst is reviewing a table that stores customer orders. The table contains columns: OrderID, CustomerName, Product1, Product1Qty, Product2, Product2Qty. This design violates which normal form?

Accepted Answer

First normal form (1NF). The table violates First Normal Form (1NF) because it contains repeating groups (Product1, Product1Qty, Product2, Product2Qty) instead of storing each product in a separate row. 1NF requires that each column contains atomic values and that there are no repeating groups or arrays. The presence of multiple product columns for a single order breaks this atomicity and normalization rule.

Answer

No violation

Answer

Third normal form (3NF)

Answer

Second normal form (2NF)

Question 17

A company is designing a data pipeline to process streaming data from social media feeds. Which THREE of the following are characteristics of streaming data? (Select THREE).

Accepted Answer

Data is unbounded and infinite. Streaming data is inherently unbounded and infinite because social media feeds generate a continuous, never-ending flow of events. Unlike batch data, there is no natural end to the stream; new tweets, posts, or interactions arrive constantly, making the dataset theoretically infinite in size.

Answer

Data is processed in micro-batches

Answer

Data is stored permanently before processing

Question 18

A data analyst is using a public API to collect historical weather data. The API has a rate limit of 100 requests per minute, but the analyst needs to retrieve 10,000 records as quickly as possible. What strategy should be used?

Accepted Answer

Paginate with appropriate delays. Paginating with appropriate delays respects the API's rate limit of 100 requests per minute while maximizing throughput. By splitting the 10,000 records into pages (e.g., 100 records per page) and sending requests at a rate just under the limit (e.g., one request every 0.6 seconds), the analyst can retrieve all data in approximately 100 minutes without triggering HTTP 429 rate-limit errors.

Answer

Increase the request rate

Answer

Use multiple API keys

Answer

Download a precompiled dataset

Question 19

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?

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 20

A data analyst is preparing a report on customer satisfaction scores. To comply with GDPR, which THREE actions must be taken? (Select THREE.)

Accepted Answer

Ensure aggregates do not identify individuals. Option B is correct because GDPR's data minimization and purpose-limitation principles require that aggregated statistics used for reporting must not allow re-identification of any individual, so ensuring aggregates do not identify individuals protects data subjects' privacy. Option D is correct because anonymizing personally identifiable information (PII) such as names, email addresses, and account numbers removes the personal data from scope of GDPR processing, satisfying the regulation's requirement to protect identifiable data. Option E is correct because GDPR Article 5(1)(e) mandates storage limitation, meaning the analyst must establish defined data retention periods for the report data rather than keeping it longer than necessary. Option A is incorrect because retaining data indefinitely violates the GDPR storage-limitation principle. Option C is incorrect because including customer names for context unnecessarily introduces identifiable personal data, contradicting data minimization and the anonymization requirement.

Answer

Retain data indefinitely for analysis

Answer

Include customer names for context