Regression metrics

Edge AI Lab

tags
edge-ai-lab

The following metrics apply to regression tasks:

Metric Description Use case
MAE Mean Absolute Error. The average absolute difference between predicted and actual values. Direction of error does not matter. When you want to minimize average prediction error.
MSE Mean Squared Error. The average of squared differences between predicted and actual values. Always non-negative, lower is better. Penalizes large errors more strongly. When large individual errors should be penalized.
Coefficient of Determination. Measures the proportion of variance in the target variable explained by the model. A value of 1 means perfect predictions, 0 means no explanatory power. When you want to know how well the model explains the variance in the data.
RMSE Root Mean Squared Error. The square root of MSE. Lower is better. When large individual errors should be penalized heavily.
RMSLE Root Mean Squared Logarithmic Error. Lower is better. Penalizes underestimates more than overestimates. When large differences between high-value predictions should be penalized less.
RMSPE Root Mean Squared Percentage Error. Measures percentage error between predicted and actual values. Lower is better. Rows with 0 in the target variable are excluded. When you need errors expressed as percentages.
Max AE Maximum Absolute Error. The largest absolute difference between predicted and actual values. When you want to understand the maximum possible deviation.
Min AE Minimum Absolute Error. The smallest absolute difference between predicted and actual values. It indicates the best-case prediction accuracy for individual samples. When you want to see how close the model can get to perfect predictions.