Checkpoint: Regression

Question 1 of 5

• The model is biased towards predicting too early.
• RMSE and MAE should be equal, so one of them was computed wrongly.
• Most predictions are close, but a few are wildly off, and squaring makes those dominate RMSE.
• The model explains 95% of the variance.

Question 2 of 5

• Add polynomial features so the model has more capacity.
• Increase the learning rate tenfold.
• Use a smaller test set.
• Standardise the features so the loss surface isn’t a long narrow valley, then rerun.

Question 3 of 5

• Train it for longer.
• Add a Ridge penalty and choose alpha on validation data.
• Report the training R², since validation sets are small.
• Raise the degree to 6 to capture the pattern fully.

Question 4 of 5

• Lasso refuses to fit features with different units.
• Nothing; Lasso is scale-invariant like decision trees.
• Income will get a huge coefficient and dominate the model.
• The penalty treats the coefficients unfairly: features with small numeric ranges need big coefficients and get shrunk or zeroed for their units, not their usefulness.

Question 5 of 5

• Its squared error on the test set equals that of always predicting the mean of y_test.
• Every prediction is exactly zero.
• The model has no coefficients.
• The model is perfect but the metric failed.