1. Question 1

What does
cross_val_predict(lr_model, X_train, y_train, cv=3)
return?

• ❌ Predicted values of the test set

• ❌ List of residual errors

• ❌ Average R² score

• ✅ Predicted values for each training point using cross-validation

Explanation:

cross_val_predict returns cross-validated predictions, not scores or residuals.

---

2. Question 2

Correct way to define a list of alpha values for Ridge regression grid search:

• ✅ parameter = [{‘alpha’: [1, 10, 100]}]

• ❌ grid = alpha:[1,10,100]

• ❌ alpha = Ridge([1, 10, 100])

• ❌ parameter = [alpha: 1, 10, 100]

Explanation:

GridSearchCV requires a dictionary inside a list with parameter name as key.

---

3. Question 3

R² = 0.99 with a 100-degree polynomial → check for overfitting by:

• ✅ Evaluate the model on the test dataset

• ❌ Reduce features first

• ❌ Use cross_val_predict on training

• ❌ Select model based only on training score

Explanation:

Overfitting is detected when training score is high but test score is low.

---

4. Question 4

Why choose Ridge Regression?

• ✅ To reduce overfitting by penalizing large coefficients

• ❌ To remove irrelevant features

• ❌ To increase flexibility

• ❌ To reduce complexity only

Explanation:

Ridge adds L2 regularization, shrinking coefficients to prevent overfitting.

---

5. Question 5

(Image: blue curve follows noise, orange is true function)

• ❌ Good fit

• ❌ No conclusion

• ✅ It displays overfitting

• ❌ Underfitting

Explanation:

The blue curve “wiggles” excessively and follows noise → classic overfitting.

---

🧾 Summary Table

---