Forecasting Models

NeuralForecast currently offers the following models.

Model1	AutoModel2	Family3	Univariate / Multivariate4	Forecast Type5	Exogenous6
Autoformer	AutoAutoformer	Transformer	Univariate	Direct	F
BiTCN	AutoBiTCN	CNN	Univariate	Direct	F/H/S
DeepAR	AutoDeepAR	RNN	Univariate	Direct	F/S
DeepNPTS	AutoDeepNPTS	MLP	Univariate	Direct	F/H/S
DilatedRNN	AutoDilatedRNN	RNN	Univariate	Direct	F/H/S
FEDformer	AutoFEDformer	Transformer	Univariate	Direct	F
GRU	AutoGRU	RNN	Univariate	Both8	F/H/S
HINT	AutoHINT	Any7	Both7	Both7	F/H/S
Informer	AutoInformer	Transformer	Univariate	Direct	F
iTransformer	AutoiTransformer	Transformer	Multivariate	Direct	-
KAN	AutoKAN	KAN	Univariate	Direct	F/H/S
LSTM	AutoLSTM	RNN	Univariate	Both8	F/H/S
MLP	AutoMLP	MLP	Univariate	Direct	F/H/S
MLPMultivariate	AutoMLPMultivariate	MLP	Multivariate	Direct	F/H/S
NBEATS	AutoNBEATS	MLP	Univariate	Direct	-
NBEATSx	AutoNBEATSx	MLP	Univariate	Direct	F/H/S
NHITS	AutoNHITS	MLP	Univariate	Direct	F/H/S
NLinear	AutoNLinear	MLP	Univariate	Direct	-
PatchTST	AutoPatchTST	Transformer	Univariate	Direct	-
RMoK	AutoRMoK	KAN	Multivariate	Direct	-
RNN	AutoRNN	RNN	Univariate	Both8	F/H/S
SOFTS	AutoSOFTS	MLP	Multivariate	Direct	-
StemGNN	AutoStemGNN	GNN	Multivariate	Direct	-
TCN	AutoTCN	CNN	Univariate	Direct	F/H/S
TFT	AutoTFT	Transformer	Univariate	Direct	F/H/S
TiDE	AutoTiDE	MLP	Univariate	Direct	F/H/S
TimeMixer	AutoTimeMixer	MLP	Multivariate	Direct	-
TimeLLM	-	LLM	Univariate	Direct	-
TimesNet	AutoTimesNet	CNN	Univariate	Direct	F
TimeXer	AutoTimeXer	Transformer	Multivariate	Direct	F
TSMixer	AutoTSMixer	MLP	Multivariate	Direct	-
TSMixerx	AutoTSMixerx	MLP	Multivariate	Direct	F/H/S
VanillaTransformer	AutoVanillaTransformer	Transformer	Univariate	Direct	F

1. Model: The model name.
2. AutoModel: NeuralForecast offers most models also in an Auto* version, in which the hyperparameters of the underlying model are automatically optimized and the best-performing model for a validation set is selected. The optimization methods include grid search, random search, and Bayesian optimization.
3. Family: The main neural network architecture underpinning the model.
4. Univariate / Multivariate: A multivariate model explicitly models the interactions between multiple time series in a dataset and will provide predictions for multiple time series concurrently. In contrast, a univariate model trained on multiple time series implicitly models interactions between multiple time series and provides predictions for single time series concurrently. Multivariate models are typically computationally expensive and empirically do not necessarily offer better forecasting performance compared to using a univariate model.
5. Forecast Type: Direct forecast models are models that produce all steps in the forecast horizon at once. In contrast, recursive forecast models predict one-step ahead, and subsequently use the prediction to compute the next step in the forecast horizon, and so forth. Direct forecast models typically suffer less from bias and variance propagation as compared to recursive forecast models, whereas recursive models can be computationally less expensive.
6. Exogenous: Whether the model accepts exogenous variables. This can be exogenous variables that contain information about the past and future (F), about the past only (historical, H), or that contain static information (static, S).
7. HINT is a modular framework that can combine any type of neural architecture with task-specialized mixture probability and advanced hierarchical reconciliation strategies.
8. Models that can produce forecasts recursively and direct. For example, the RNN model uses an RNN to encode the past sequence, and subsequently the user can choose between producing forecasts recursively using the RNN or direct using an MLP that uses the encoded sequence as input. The models feature an recursive=False feature that sets how they produce forecasts.