Regression performance quantifies the capability of a regression model to make predictions that are close to the true values. In machine learning, regression performance entails applying a trained model to test or production data and evaluating the performance of the predicted results against known outcomes.
Similar to evaluating performance for other supervised machine learning approaches such as classification, quantifying regression model performance is key to selecting best models for AI applications and monitoring them over time. Evaluating performance of a regression model, though, requires a different approach and different metrics from evaluating classification models [Machine Learning for Managers].
Example [Machine Learning for Managers]:
MAE measures the absolute error between predicted and observed values. For example, an MAE value of 0 indicates there is no difference between predicted values and observed values. In practice, MAE is a popular error metric because it is both intuitive and easy to compute.
Mathematically, MAE is defined as:
Where n is the total number of evaluated samples, xi is the observed (actual) output, and yi is the predicted output.
C3.ai offers out-of-the-box regression-based machine learning pipes built around open-source libraries like sklearn to enable calculation of regression metrics like mean absolute error (MAE), mean absolute percentage error (MAPE) and root mean square error (RMSE).