WGU D492 Data Analytics - Applications | proctored CompTIA
Data+ DA0-002 certification exam study guide | Questions and
Answers | 2026 Update | 100% Correct - 117 Questions and
Answers Already Graded A+ Premium Exam Tested And
Verified
Subject Area Data Analytics
Description This exam assesses advanced knowledge and application of data analytics
concepts aligned with the CompTIA Data+ (DA0-002) certification. Topics
include data governance, quality, analysis processes, statistics, regression,
clustering, visualization, and data warehousing. Students must demonstrate
multi-step reasoning and synthesis across domains.
Expected Grade A+
Total Questions 117
Duration 3 hours
Learning Outcomes 1. Apply appropriate data governance and quality frameworks to ensure data
integrity and compliance.
2. Select and justify statistical methods for hypothesis testing and predictive
modeling.
3. Design and evaluate data visualizations that accurately communicate insights to
stakeholders.
4. Implement ETL processes and data warehousing strategies for efficient data
management.
Accreditation US university (Ivy League / R1 research level) standards; aligns with CompTIA
Data+ DA0-002 objectives.
Page 1
,1. A healthcare organization collects patient data for treatment but also uses it for
marketing without explicit consent. Which data governance principle is most directly
violated?
Answer: Data minimization
Data minimization requires collecting only necessary data for a specified purpose; using
data for marketing without consent violates purpose limitation and minimization. Data
provenance is about origin tracking, transparency about disclosure, stewardship about
management roles.
2. A dataset contains 10,000 customer records. 20% of 'age' values are missing, and
5% of 'income' values are outliers due to data entry errors. Which sequence of data
quality processes is most appropriate?
Answer: Assign mean age to missing values, then cap outliers at 99th percentile
With 20% missing on age, deletion would lose too much data; mean imputation is
acceptable for approximately normal distributions. For income outliers, capping at a
percentile reduces bias without losing data. Option B loses data unnecessarily; D's log
transform alters interpretability.
3. In the CRISP-DM framework, after building a predictive model, the team
discovers the business problem was incorrectly defined. Which phase should be
revisited?
Answer: Business understanding
CRISP-DM is iterative; if the business problem is misdefined, the root cause lies in the
Business Understanding phase. Revisiting that phase allows redefinition before
re-entering subsequent phases. The Evaluation phase checks if the model meets
business objectives, but it does not correct the problem definition itself.
4. A marketing analyst wants to cluster customers based on purchase frequency,
average transaction value, and recency. Euclidean distance is used but yields poor
separation. The correct deficiency is that Euclidean distance:
Answer: Is sensitive to the scale of measurement
Euclidean distance is sensitive to the scale of variables; variables measured in different
units (e.g., frequency counts vs. dollar amounts) dominate the distance calculation.
Standardization (e.g., z-scores) is typically required before clustering. The other options
are not inherent flaws for this scenario.
Page 2
,5. A data scientist runs a linear regression with 5 predictors. The model has an
R-squared of 0.95 but most predictors are not statistically significant. The most
likely issue is:
Answer: Multicollinearity
High R-squared with insignificant predictors strongly suggests multicollinearity, where
predictors are correlated, inflating standard errors and reducing t-statistics.
Overfitting would typically yield high R-squared but poor prediction;
heteroscedasticity affects inference but not necessarily significance pattern.
6. A data analyst wants to compare the means of two independent groups with
non-normal distributions and small sample sizes (n1=12, n2=14). Which test is most
appropriate?
Answer: Mann-Whitney U test
The Mann-Whitney U test is a non-parametric alternative to the independent t-test,
suitable for non-normal distributions and small samples. Paired t-test is for dependent
groups; ANOVA is for >2 groups; chi-square tests categorical associations.
7. A data visualization designer must choose a chart to show the relationship
between customer satisfaction score (1-10) and time spent on support call (minutes).
Both variables are continuous. The best choice is:
Answer: Scatter plot with trend line
A scatter plot effectively displays the relationship between two continuous variables,
and a trend line helps assess correlation and direction. Bar charts are for categorical
data, pie charts for parts of a whole, box plots for distribution of one variable.
8. In an ETL pipeline for a data warehouse, the extract phase reads from a
transactional database that updates continuously. To ensure consistent snapshots for
reporting, the best approach is:
Answer: Change data capture with timestamp-based extraction
Change data capture (CDC) with timestamps captures only modified records, ensuring
consistency and efficiency. Full refresh is heavy; real-time trickle-feed may introduce
inconsistency if not managed; extracting only last row loses data. CDC preserves state
at a point in time.
Page 3
, 9. A company stores raw transaction logs as JSON files. An analyst needs to query
these for aggregation without a predefined schema. Which technology is most
appropriate?
Answer: Data lake with schema-on-read
JSON files with no predefined schema are best suited for a data lake using
schema-on-read, where the schema is applied at query time. Relational databases
require predefined schemas; operational stores are for OLTP; data marts are for
specific business units and typically structured.
10. A data analyst uses k-means clustering with k=4 on a dataset of 10,000 points.
After running, 2 clusters contain 1 point each. The most likely cause is:
Answer: Outliers in the data
Very small clusters in k-means often indicate outliers: points far from others become
separate clusters. Improper initialization could cause empty clusters but not singleton
ones; too many iterations wouldn't cause this; Euclidean on categorical data would
cause other issues but not typically singleton clusters.
11. A data analyst finds that customer birth dates in a CRM database are all
recorded as January 1, 1900, for entries where the actual birth date is unknown.
Which data quality dimension is most directly compromised?
Answer: Accuracy
Accuracy refers to how well data reflects reality. Entering a placeholder date when the
true value is unknown introduces inaccuracy, not just incompleteness. Validity relates
to format, completeness to presence of data, and timeliness to age of data.
12. In the CRISP-DM framework, which phase directly follows 'Data Preparation'
and involves selecting modeling techniques and building models?
Answer: Modeling
CRISP-DM phases: Business Understanding, Data Understanding, Data Preparation,
Modeling, Evaluation, Deployment. Modeling comes after Data Preparation and before
Evaluation.
Page 4
Data+ DA0-002 certification exam study guide | Questions and
Answers | 2026 Update | 100% Correct - 117 Questions and
Answers Already Graded A+ Premium Exam Tested And
Verified
Subject Area Data Analytics
Description This exam assesses advanced knowledge and application of data analytics
concepts aligned with the CompTIA Data+ (DA0-002) certification. Topics
include data governance, quality, analysis processes, statistics, regression,
clustering, visualization, and data warehousing. Students must demonstrate
multi-step reasoning and synthesis across domains.
Expected Grade A+
Total Questions 117
Duration 3 hours
Learning Outcomes 1. Apply appropriate data governance and quality frameworks to ensure data
integrity and compliance.
2. Select and justify statistical methods for hypothesis testing and predictive
modeling.
3. Design and evaluate data visualizations that accurately communicate insights to
stakeholders.
4. Implement ETL processes and data warehousing strategies for efficient data
management.
Accreditation US university (Ivy League / R1 research level) standards; aligns with CompTIA
Data+ DA0-002 objectives.
Page 1
,1. A healthcare organization collects patient data for treatment but also uses it for
marketing without explicit consent. Which data governance principle is most directly
violated?
Answer: Data minimization
Data minimization requires collecting only necessary data for a specified purpose; using
data for marketing without consent violates purpose limitation and minimization. Data
provenance is about origin tracking, transparency about disclosure, stewardship about
management roles.
2. A dataset contains 10,000 customer records. 20% of 'age' values are missing, and
5% of 'income' values are outliers due to data entry errors. Which sequence of data
quality processes is most appropriate?
Answer: Assign mean age to missing values, then cap outliers at 99th percentile
With 20% missing on age, deletion would lose too much data; mean imputation is
acceptable for approximately normal distributions. For income outliers, capping at a
percentile reduces bias without losing data. Option B loses data unnecessarily; D's log
transform alters interpretability.
3. In the CRISP-DM framework, after building a predictive model, the team
discovers the business problem was incorrectly defined. Which phase should be
revisited?
Answer: Business understanding
CRISP-DM is iterative; if the business problem is misdefined, the root cause lies in the
Business Understanding phase. Revisiting that phase allows redefinition before
re-entering subsequent phases. The Evaluation phase checks if the model meets
business objectives, but it does not correct the problem definition itself.
4. A marketing analyst wants to cluster customers based on purchase frequency,
average transaction value, and recency. Euclidean distance is used but yields poor
separation. The correct deficiency is that Euclidean distance:
Answer: Is sensitive to the scale of measurement
Euclidean distance is sensitive to the scale of variables; variables measured in different
units (e.g., frequency counts vs. dollar amounts) dominate the distance calculation.
Standardization (e.g., z-scores) is typically required before clustering. The other options
are not inherent flaws for this scenario.
Page 2
,5. A data scientist runs a linear regression with 5 predictors. The model has an
R-squared of 0.95 but most predictors are not statistically significant. The most
likely issue is:
Answer: Multicollinearity
High R-squared with insignificant predictors strongly suggests multicollinearity, where
predictors are correlated, inflating standard errors and reducing t-statistics.
Overfitting would typically yield high R-squared but poor prediction;
heteroscedasticity affects inference but not necessarily significance pattern.
6. A data analyst wants to compare the means of two independent groups with
non-normal distributions and small sample sizes (n1=12, n2=14). Which test is most
appropriate?
Answer: Mann-Whitney U test
The Mann-Whitney U test is a non-parametric alternative to the independent t-test,
suitable for non-normal distributions and small samples. Paired t-test is for dependent
groups; ANOVA is for >2 groups; chi-square tests categorical associations.
7. A data visualization designer must choose a chart to show the relationship
between customer satisfaction score (1-10) and time spent on support call (minutes).
Both variables are continuous. The best choice is:
Answer: Scatter plot with trend line
A scatter plot effectively displays the relationship between two continuous variables,
and a trend line helps assess correlation and direction. Bar charts are for categorical
data, pie charts for parts of a whole, box plots for distribution of one variable.
8. In an ETL pipeline for a data warehouse, the extract phase reads from a
transactional database that updates continuously. To ensure consistent snapshots for
reporting, the best approach is:
Answer: Change data capture with timestamp-based extraction
Change data capture (CDC) with timestamps captures only modified records, ensuring
consistency and efficiency. Full refresh is heavy; real-time trickle-feed may introduce
inconsistency if not managed; extracting only last row loses data. CDC preserves state
at a point in time.
Page 3
, 9. A company stores raw transaction logs as JSON files. An analyst needs to query
these for aggregation without a predefined schema. Which technology is most
appropriate?
Answer: Data lake with schema-on-read
JSON files with no predefined schema are best suited for a data lake using
schema-on-read, where the schema is applied at query time. Relational databases
require predefined schemas; operational stores are for OLTP; data marts are for
specific business units and typically structured.
10. A data analyst uses k-means clustering with k=4 on a dataset of 10,000 points.
After running, 2 clusters contain 1 point each. The most likely cause is:
Answer: Outliers in the data
Very small clusters in k-means often indicate outliers: points far from others become
separate clusters. Improper initialization could cause empty clusters but not singleton
ones; too many iterations wouldn't cause this; Euclidean on categorical data would
cause other issues but not typically singleton clusters.
11. A data analyst finds that customer birth dates in a CRM database are all
recorded as January 1, 1900, for entries where the actual birth date is unknown.
Which data quality dimension is most directly compromised?
Answer: Accuracy
Accuracy refers to how well data reflects reality. Entering a placeholder date when the
true value is unknown introduces inaccuracy, not just incompleteness. Validity relates
to format, completeness to presence of data, and timeliness to age of data.
12. In the CRISP-DM framework, which phase directly follows 'Data Preparation'
and involves selecting modeling techniques and building models?
Answer: Modeling
CRISP-DM phases: Business Understanding, Data Understanding, Data Preparation,
Modeling, Evaluation, Deployment. Modeling comes after Data Preparation and before
Evaluation.
Page 4
