Regression: objectives and metrics


Name	Used for optimization	User-defined parameters	Formula and/or description
MAE	+	use_weights Default: true	Calculation principles $\text{[math]}$
MAPE	+	use_weights Default: true	Calculation principles $\text{[math]}$
Poisson	+	–	Calculation principles $\text{[math]}$
Quantile	+	alpha Default: 0.5	Calculation principles $\text{[math]}$
RMSE	+	–	Calculation principles $\text{[math]}$
LogLinQuantile	+	alpha Default: 0.5	Calculation principles Depends on the condition for the ratio of the label value and the resulting value: $\text{[math]}$
Lq	+	q Default: Obligatory parameter	Calculation principles $\text{[math]}$
Huber	+	delta Default: Obligatory parameter use_weights Default: true	Calculation principles $\text{[math]}$ $\text{[math]}$
Expectile	+	use_weights Default: true alpha Default: 0.5	Calculation principles $\text{[math]}$
FairLoss	–	use_weights Default: true smoothness Default: 1.0	Calculation principles $\text{[math]}$ $\text{[math]}$
NumErrors	–	greater_than use_weights Default: true	The proportion of predictions, for which the difference from the label value exceeds the specified value `greater_than`. Calculation principles $\text{[math]}$ $\text{[math]}$
SMAPE	–	use_weights Default: true	Calculation principles $\text{[math]}$
R2	–	use_weights Default: true	Calculation principles $\text{[math]}$ $\text{[math]}$ is the average label value: $\text{[math]}$
MSLE	–	use_weights Default: true	Calculation principles $\text{[math]}$
MedianAbsoluteError	–	–	Calculation principles $\text{[math]}$


Name	Used for optimization	User-defined parameters	Formula and/or description
MAE	+	use_weights Default: true	Calculation principles $\text{[math]}$
MAPE	+	use_weights Default: true	Calculation principles $\text{[math]}$
Poisson	+	–	Calculation principles $\text{[math]}$
Quantile	+	alpha Default: 0.5	Calculation principles $\text{[math]}$
RMSE	+	–	Calculation principles $\text{[math]}$
LogLinQuantile	+	alpha Default: 0.5	Calculation principles Depends on the condition for the ratio of the label value and the resulting value: $\text{[math]}$
Lq	+	q Default: Obligatory parameter	Calculation principles $\text{[math]}$
Huber	+	delta Default: Obligatory parameter use_weights Default: true	Calculation principles $\text{[math]}$ $\text{[math]}$
Expectile	+	use_weights Default: true alpha Default: 0.5	Calculation principles $\text{[math]}$
FairLoss	–	use_weights Default: true smoothness Default: 1.0	Calculation principles $\text{[math]}$ $\text{[math]}$
NumErrors	–	greater_than use_weights Default: true	The proportion of predictions, for which the difference from the label value exceeds the specified value `greater_than`. Calculation principles $\text{[math]}$ $\text{[math]}$
SMAPE	–	use_weights Default: true	Calculation principles $\text{[math]}$
R2	–	use_weights Default: true	Calculation principles $\text{[math]}$ $\text{[math]}$ is the average label value: $\text{[math]}$
MSLE	–	use_weights Default: true	Calculation principles $\text{[math]}$
MedianAbsoluteError	–	–	Calculation principles $\text{[math]}$