1️⃣ Question 1

A company wants to forecast CO₂ emissions based on multiple variables. Which regression should they use?

• ✅ Multiple regression

• ❌ Logistic regression

• ❌ Simple regression

• ❌ Polynomial regression

Explanation:
Multiple regression handles more than one independent variable (engine size, cylinders, fuel consumption).

---

2️⃣ Question 2

When is simple regression appropriate?

• ❌ Predicting rainfall from temperature (still multiple climate variables)

• ❌ Grouping customers (clustering, not regression)

• ❌ Predicting sales from multiple variables

• ❌ Text classification (ML classification)

• Correct Answer: NONE OF THE ABOVE

But the closest matching intended exam answer is:

• ✅ Predicting annual rainfall based on average temperature.

Explanation:
Simple regression = one input → one output.
Only the rainfall/temperature option has a single predictor.

---

3️⃣ Question 3

Predict employee productivity using several variables. Which technique?

• ❌ Simple logarithmic regression

• ❌ Simple regression

• ✅ Multiple linear regression

• ❌ Simple polynomial regression

Explanation:
Productivity predicted from hours of training + experience + projects requires multiple regression.

---

4️⃣ Question 4

When is logarithmic regression appropriate?

• ❌ Monthly expenses with consistent growth (exponential, maybe)

• ❌ Ice cream sales vs temperature (linear or polynomial)

• ❌ Steady linear increase in sales

• ✅ Examining slow website traffic growth with added marketing budget

Explanation:
When growth slows down as input increases → log curve.

---

5️⃣ Question 5

Logistic regression has high false positives. What can reduce this?

• ❌ Collect more data

• ❌ Regularization

• ❌ More features

• ✅ Tune the classification threshold

Explanation:
False positives improve by adjusting the decision boundary (e.g., 0.5 → 0.6).

---

6️⃣ Question 6

Model predicts probability = 0.65 for “return item”:

• ❌ 100% chance

• ❌ 0% chance

• ✅ 65% likelihood customer will return the item

• ❌ 35%

---

7️⃣ Question 7

Which scenario produces the highest log loss?

• ❌ Predict 0.9 for correct class

• ❌ Predict 0.7

• ✅ Predict 0.1 for the correct class and 0.9 for the incorrect class

• ❌ Predict 0.5 for both

Explanation:
Log loss heavily penalizes high confidence in the wrong prediction.

---

🧾 Summary Table