Optimization Objectives

NeuralForecast is a highly modular framework capable of augmenting a wide variety of robust neural network architectures with different point or probability outputs as defined by their optimization objectives.

Point losses

Scale-Dependent	Percentage-Errors	Scale-Independent	Robust
MAE	MAPE	MASE	Huber
MSE	sMAPE		Tukey
RMSE			HuberMQLoss

Probabilistic losses

Parametric Probabilities	Non-Parametric Probabilities
Normal	QuantileLoss
StudenT	MQLoss
Poisson	HuberQLoss
Negative Binomial	HuberMQLoss
Tweedie	IQLoss
PMM	HuberIQLoss
GMM	ISQF
NBMM

Forecasting Models Exogenous Variables

⌘I

​Point losses

​Probabilistic losses

Point losses

Probabilistic losses