Free SAS Certified Associate: Applied Statistics for Machine Learning Practice Questions

Sample Practice Questions

Try these 10 sample questions from the SAS Certified Associate: Applied Statistics for Machine Learning practice bank. Each includes the answer and explanation. Continue with the interactive practice above to review the full question set.

1. 1Which big-data dimension concerns the amount of information that must be stored and processed?

   1. A. Variety
   2. B. Volume
   3. C. Velocity
   4. D. Veracity

   Answer: B

   Explanation: The SAS course states, "Volume: Volume refers to the amount of data from terabytes to petabytes and up." Volume concerns quantity; variety concerns formats, velocity concerns speed, and veracity concerns quality.

   Reference: Statistics You Need to Know for Machine Learning — course text — What Is Big Data? — Volume; derived line 453

2. 2An analyst wants to generalize results to every active account but analyzes records for only 800 selected accounts. What do the 800 records represent?

   1. A. The population of interest
   2. B. The population parameter
   3. C. The sampling distribution
   4. D. The sample available for analysis

   Answer: D

   Explanation: SAS defines a sample as "A subset of the population that is available for analysis." All active accounts form the population of interest; the selected records constitute its sample, rather than a numeric parameter or a distribution of a statistic.

   Reference: SAS Enterprise Miner Extension Nodes: Developer’s Guide — Terminology — Terminology — Population; Sample

3. 3What does a Pearson correlation of -0.92 indicate about two continuous variables?

   1. A. A weak negative linear association
   2. B. A strong negative linear association
   3. C. A strong positive linear association
   4. D. A weak positive linear association

   Answer: B

   Explanation: The course states, "Stronger relationships are closer to +1 or -1". A value of -0.92 is close to -1, so the linear association is strong and negative; the minus sign specifies direction, not weakness.

   Reference: Statistics You Need to Know for Machine Learning — course text — Pearson Correlation — feedback; derived lines 3078–3084

4. 4Which objective most directly characterizes the purpose of predictive modeling?

   1. A. Testing every explanatory coefficient for statistical significance
   2. B. Selecting the equation that gives the strongest causal interpretation
   3. C. Estimating causal effects from associations without further assumptions
   4. D. Producing accurate outcomes for new or future observations

   Answer: D

   Explanation: The course states, "الهدف الأساسي هو دقة التنبؤ." Predictive modeling prioritizes prediction accuracy for new or future observations, whereas explanatory modeling focuses on understanding and testing relationships. Predictive accuracy alone does not establish causality.

   Reference: Statistics You Need to Know for Machine Learning — course text — To Explain or Predict? — Predictive modeling goal; derived lines 4200–4220

5. 5Which task is supervised regression rather than supervised classification?

   1. A. Predicting a machine's operating temperature from labelled historical records
   2. B. Predicting which of three fault categories a machine will enter
   3. C. Predicting whether a machine will fail during its next shift
   4. D. Predicting whether an inspection result will be pass or fail

   Answer: A

   Explanation: The course states, "Regression analysis is supervised prediction where the target is a continuous variable." Temperature is a continuous target; fault categories, failure indicators, and pass/fail outcomes are categorical targets for classification.

   Reference: Statistics You Need to Know for Machine Learning — course text — Analytical Methods for Predictive Modeling — Regression and classification; derived lines 4325–4331

6. 6Which requirement is driven most directly by the velocity dimension of big data?

   1. A. Reconciling inconsistent values collected from multiple sources
   2. B. Interpreting a mixture of numeric tables and recorded audio
   3. C. Provisioning capacity to store petabytes of historical records
   4. D. Analyzing incoming sensor readings within seconds of their arrival

   Answer: D

   Explanation: SAS states that "data streams into businesses at an unprecedented speed and must be handled in a timely manner." Processing incoming readings within seconds addresses velocity. Reconciling inconsistent values addresses veracity, handling mixed formats addresses variety, and storage quantity addresses volume.

   Reference: Big Data: What it is and why it matters — History of Big Data — Velocity

7. 7A researcher uses an average calculated from surveyed households to estimate the average for all households in a city. Which description is correct?

   1. A. A population parameter estimates a sample statistic.
   2. B. A sample statistic estimates a population parameter.
   3. C. A population parameter estimates a sampling frame.
   4. D. A sampling frame estimates a sample statistic.

   Answer: B

   Explanation: The SAS course explains, "Population parameters are generally unknown and can be estimated or tested using sample statistics." The surveyed-household average is a statistic; the citywide average is the parameter being estimated.

   Reference: Statistics You Need to Know for Machine Learning — course text — Fundamental Statistical Concepts — Lesson summary; derived line 2875

8. 8A scatterplot shows a strong U-shaped relationship, but Pearson correlation is close to zero. Which conclusion is justified?

   1. A. The two variables must be statistically independent.
   2. B. The U shape establishes a strong negative linear relationship.
   3. C. The U shape establishes a strong positive linear relationship.
   4. D. A strong nonlinear relationship can coexist with near-zero linear correlation.

   Answer: D

   Explanation: The course states that Pearson r "measures the strength of the linear association between continuous variables". A near-zero value describes weak linear association, while the U-shaped scatterplot shows a nonlinear pattern. Neither a positive nor a negative straight-line characterization captures that curve, and zero linear correlation does not establish independence.

   Reference: Statistics You Need to Know for Machine Learning — course text — Pearson Correlation — feedback; Nonlinear relationships; derived lines 3030–3040 and 3078–3084

9. 9After all model fitting and tuning are complete, which data partition should be used once to estimate final generalization performance?

   1. A. The training partition used to estimate coefficients
   2. B. The validation partition repeatedly used to choose models
   3. C. An untouched test partition reserved for final assessment
   4. D. The combined training and validation partitions used during development

   Answer: C

   Explanation: The course states, "The test data set has only one use: to give a final honest estimate of generalization." Data used in fitting or repeated model choice do not provide the same untouched final assessment.

   Reference: Statistics You Need to Know for Machine Learning — course text — Honest Assessment — feedback; derived lines 4428–4434

10. 10An analyst has customer attributes but no outcome labels and wants to discover groups of similar customers. Which learning approach matches the task?

    1. A. Supervised classification
    2. B. Reinforcement learning
    3. C. Supervised regression
    4. D. Unsupervised learning

    Answer: D

    Explanation: The course states, "unsupervised learning discovers underlying data structures such as clusters". Grouping customers from their input similarities without outcome labels is unsupervised learning, rather than predicting labelled outcomes or learning from action rewards.

    Reference: Statistics You Need to Know for Machine Learning — course text — Machine Learning Approaches — feedback; derived lines 5412–5420