03
Domain 03
Regression & Regularization
Reading coefficients, log targets, ridge and lasso, choosing between RMSE, MAE, and MAPE, and what residuals say.
6 questions
- Q13How should the coefficient on garage_cars be read?A $9,800 coefficient on garage spaces. What it says — and what it doesn't.
- Q14What are the MAE and the RMSE, and what does the difference tell you?Five errors, one big miss. Compute MAE and RMSE and see them disagree.
- Q15What is the most appropriate fix?Residuals that fan out from $8K stores to $2M stores. The classic fix.
- Q16Why does lasso produce exact zeros while ridge does not?176 of 200 coefficients are exactly zero. Only one penalty can do that.
- Q17Which statement is correct?Two models, RMSE and R² on the table. Do the metrics agree?
- Q18What is each product's absolute percentage error, and what does this expose about MAPE?A 10-unit miss shows up as 500% error. MAPE's dirty secret.