Machine Learning
Mean Squared Error, Root Mean Squared Error, Mean Absolute Error, R-squared are metrics for evaluation which type of models?
Regression models.
Card 1 of 259. Answer: Regression models.
Regression models.
Card 1 of 259. Answer: Regression models.
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
Regression models, which predict a continuous number such as a price or a delivery time. MAE, MSE and RMSE measure the typical size of the prediction errors, so lower is better. R-squared compares the model with a baseline that always predicts the mean, so higher is better.
Example: actual values 10, 12, 14 and 20 with predictions 11, 11, 15 and 18 give errors of −1, 1, −1 and 2. MAE = 5/4 = 1.25. MSE = 7/4 = 1.75, and RMSE = √1.75 ≈ 1.32, in the target's units. The actuals average 14, so the total sum of squares is 16 + 4 + 0 + 36 = 56 and R² = 1 − 7/56 = 0.875.
Interviewers ask which to report. MAE weighs every unit of error equally, RMSE punishes larger misses more, and R² is unitless, so it reads the same at any scale. Classifiers use accuracy, precision, recall or AUC instead. One overlap: the Brier score is MSE on predicted probabilities, used to evaluate classifiers.