MGT 6203 (Data Analytics in Business) Final Exam
Practice Questions and Answers
Question 1: Regression
Q: A company wants to predict monthly sales based on advertising spend. The
regression output shows:
• Intercept = 50
• Advertising coefficient = 4
•
• Regression equation:
• Interpretation: For every extra unit of advertising spend, sales increase by 4
units.
• High (0.85) means the model explains 85% of the variation in sales.
Question 2: Hypothesis Testing
Q: A retailer tests whether average customer spend is greater than $100. Sample mean
= $105, p-value = 0.03.
• Null hypothesis:
• Alternative:
• Since p-value (0.03) < 0.05, reject .
• Conclusion: Evidence suggests customers spend more than $100.
Question 3: Clustering
Q: Why might a business use k-means clustering on customer data?
• To group customers into segments based on behavior (e.g., high spenders vs.
occasional buyers).
• Helps tailor marketing strategies to each segment.
Q4. Logistic Regression
A company wants to predict whether a customer will churn (Yes/No) based on monthly
usage. The logistic regression output shows:
• Coefficient for usage = -0.05
• p-value = 0.01
• Interpretation: Higher usage decreases the probability of churn.
• Since p-value < 0.05, the effect is statistically significant.
Q5. Decision Trees
Why might a business prefer decision trees over logistic regression?
, • Decision trees are easier to interpret and visualize.
• They can capture non-linear relationships and interactions automatically.
Q6. Hypothesis Testing
A sample of 40 customers has an average spend of $120, with a standard deviation of
$15. Test whether the mean spend differs from $100 at the 5% significance level.
• Null: H_0:\mu =100
• Test statistic: t=\frac{120-100}{15/\sqrt{40}}\approx 8.44
• p-value < 0.001 → Reject H_0.
• Conclusion: Mean spend is significantly greater than $100.
Q7. Multiple Regression
A company predicts revenue using advertising spend and number of salespeople.
Regression output:
• Intercept = 20
• Advertising coefficient = 3 (p = 0.02)
• Salespeople coefficient = 5 (p = 0.10)
• Advertising significantly increases revenue (p < 0.05).
• Salespeople effect is positive but not statistically significant (p = 0.10).
• Equation: .
Q8. Correlation vs. Causation
Why is correlation not proof of causation?
• Correlation shows association, but causation requires evidence of direction and
mechanism.
• Example: Ice cream sales and drowning incidents correlate, but both are driven
by hot weather.
Q9. Hypothesis Testing
Practice Questions and Answers
Question 1: Regression
Q: A company wants to predict monthly sales based on advertising spend. The
regression output shows:
• Intercept = 50
• Advertising coefficient = 4
•
• Regression equation:
• Interpretation: For every extra unit of advertising spend, sales increase by 4
units.
• High (0.85) means the model explains 85% of the variation in sales.
Question 2: Hypothesis Testing
Q: A retailer tests whether average customer spend is greater than $100. Sample mean
= $105, p-value = 0.03.
• Null hypothesis:
• Alternative:
• Since p-value (0.03) < 0.05, reject .
• Conclusion: Evidence suggests customers spend more than $100.
Question 3: Clustering
Q: Why might a business use k-means clustering on customer data?
• To group customers into segments based on behavior (e.g., high spenders vs.
occasional buyers).
• Helps tailor marketing strategies to each segment.
Q4. Logistic Regression
A company wants to predict whether a customer will churn (Yes/No) based on monthly
usage. The logistic regression output shows:
• Coefficient for usage = -0.05
• p-value = 0.01
• Interpretation: Higher usage decreases the probability of churn.
• Since p-value < 0.05, the effect is statistically significant.
Q5. Decision Trees
Why might a business prefer decision trees over logistic regression?
, • Decision trees are easier to interpret and visualize.
• They can capture non-linear relationships and interactions automatically.
Q6. Hypothesis Testing
A sample of 40 customers has an average spend of $120, with a standard deviation of
$15. Test whether the mean spend differs from $100 at the 5% significance level.
• Null: H_0:\mu =100
• Test statistic: t=\frac{120-100}{15/\sqrt{40}}\approx 8.44
• p-value < 0.001 → Reject H_0.
• Conclusion: Mean spend is significantly greater than $100.
Q7. Multiple Regression
A company predicts revenue using advertising spend and number of salespeople.
Regression output:
• Intercept = 20
• Advertising coefficient = 3 (p = 0.02)
• Salespeople coefficient = 5 (p = 0.10)
• Advertising significantly increases revenue (p < 0.05).
• Salespeople effect is positive but not statistically significant (p = 0.10).
• Equation: .
Q8. Correlation vs. Causation
Why is correlation not proof of causation?
• Correlation shows association, but causation requires evidence of direction and
mechanism.
• Example: Ice cream sales and drowning incidents correlate, but both are driven
by hot weather.
Q9. Hypothesis Testing
