Evaluation Metrics for Regression Problems: Quick examples + Reference

Evaluation Metrics for Regression Problems: Quick examples + Reference

Last updated:
Table of Contents

WIP Alert This is a work in progress. Current information is correct but more content may be added in the future.

For evaluation metrics to be used in classification problems, see instead: Evaluation Metrics for Classification Problems: Quick Examples + References

RMSE

TODO

MAE (Mean Squared Error)

TODO

R2 (r-squared)

R2 is one of several goodnes-of-fit measures.

It measures how much of the variance of the target variable is explainable by the trained model.

$$ R^2 = \frac{\text{Variance (of target variable) which can be explained by the model}}{\text{total variance of the target variable}} $$

  • R-squared ranges between 0 (model doesn't explain any variance) and 1 (all variance of the target variable is explainable by the model)

  • Biased models (i.e. ones that consistently err "up" or "down") can still have high R-squared values!

MAPE (Mean absolute percentage error)

Median Absolute Error

TODO http://scikit-learn.org/stable/modules/model_evaluation.html#median-absolute-error

Spearman's Rho

TODO