v1.0

nbs/examples/models_intro.qmd

+---
+title: "NeuralForecast's Contents"
+---
+
+
+## Automatic Forecasting Models
+Automatic forecasting tools optimize the hyperparameters of a given model class and select the best-performing model for a validation set. The optimization methods include grid search, random search, and Bayesian optimization.
+
+| MLP-Based                                 |RNN-Based                                | Transformers                                 | CNN-Based                                   | Multivariate                               |
+|:------------------------------------------|:-------------------------------------   |:---------------------------------------------|:--------------------------------------------| :-------------------------------------------|
+|[`AutoMLP`](../models.html#automlp)        |[`AutoRNN`](../models.html#autornn)      |[`AutoTFT`](../models.html#autotft)           |[`AutoTimesNet`](../models.html#autotimesnet)| [`AutoStemGNN`](../models.html#autostemgnn) |
+|[`AutoNBEATS`](../models.html#autonbeats)  |[`AutoLSTM`](../models.html#autolstm)    |[`AutoInformer`](../models.html#autoinformer) |                                             | [`AutoHINT`](../models.html#autohint)       |
+|[`AutoNBEATSx`](../models.html#autonbeatsx)|[`AutoGRU`](../models.html#autogru)      |[`Autoformer`](../models.html#autoautoformer) |                                             |                                             |
+|[`AutoNHITS`](../models.html#autonhits)    |[`AutoTCN`](../models.html#autotcn)      |[`AutoPatchTST`](../models.html#autopatchtst) |                                             |                                             |
+|                                           |[`AutoDeepAR`](../models.html#autodeepar)|                                              |                                             |                                             |
+: {tbl-colwidths="[25,25]"}
+
+## 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.
+
+| Scale-Dependent                                              | Percentage-Errors                                                     | Scale-Independent                                              | Robust                                                 |
+|:-------------------------------------------------------------|:----------------------------------------------------------------------|:---------------------------------------------------------------|:-------------------------------------------------------|
+|[`MAE`](../losses.pytorch.html#mean-absolute-error-mae)       |[`MAPE`](../losses.pytorch.html#mean-absolute-percentage-error-mape)   |[`MASE`](../losses.pytorch.html#mean-absolute-scaled-error-mase)|[`Huber`](../losses.pytorch.html#huber-loss)            |
+|[`MSE`](../losses.pytorch.html#mean-squared-error-mse)        |[`sMAPE`](../losses.pytorch.html#symmetric-mape-smape)                 |                                                                |[`Tukey`](../losses.pytorch.html#tukey-loss)            |
+|[`RMSE`](../losses.pytorch.html#root-mean-squared-error-rmse) |                                                                       |                                                                |[`HuberMQLoss`](../losses.pytorch.html#huberized-mqloss)|
+: {tbl-colwidths="[25,25]"}
+
+|Parametric Probabilities                                      | Non-Parametric Probabilities                                 |
+|:-------------------------------------------------------------|:-------------------------------------------------------------|
+|[`Normal`](../losses.pytorch.html#distributionloss)           |[`QuantileLoss`](../losses.pytorch.html#quantile-loss)        |
+|[`StudenT`](../losses.pytorch.html#distributionloss)          |[`MQLoss`](../losses.pytorch.html#multi-quantile-loss-mqloss) |
+|[`Poisson`](../losses.pytorch.html#distributionloss)          |[`HuberQLoss`](../losses.pytorch.html#huberized-quantile-loss)|
+|[`Negative Binomial`](../losses.pytorch.html#distributionloss)|[`HuberMQLoss`](../losses.pytorch.html#huberized-mqloss)      |
+|[`Tweedie`](../losses.pytorch.html#distributionloss)          |                                                              |
+|[`PMM`](../losses.pytorch.html#poisson-mixture-mesh-pmm) /[`GMM`](../losses.pytorch.html#gaussian-mixture-mesh-gmm)  |       |
+: {tbl-colwidths="[25,25]"}
+
+## MLP-Based Model Family
+The MLP-based family operates like a classic autoencoder. Its initial layers encode raw autoregressive window into a representation, and the decoder produces the desired output based on the horizon, probability output, or point objective. Recent architectures include modifications like residual learning techniques and task-specific changes.
+
+|Model                                     | Point Forecast | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values  |
+|:-----------------------------------------|:--------------:|:----------------------:|:----------------------:|:----------------------------:|
+|[`MLP`](../models.mlp.html)               |✅              |✅                      |✅                      |✅                            |
+|[`NBEATS`](../models.nbeats.html)         |✅              |✅                      |✅                      |✅                            |
+|[`NBEATSx`](../models.nbeatsx.html)       |✅              |✅                      |✅                      |✅                            |
+|[`NHITS`](../models.nhits.html)           |✅              |✅                      |✅                      |✅                            |
+: {tbl-colwidths="[25,25]"}
+
+## RNN-Based Model Family
+The RNN-based family attempts to leverage the data's temporal structure while reducing MLPs over parametrization. Recurrent networks are dynamic and can handle sequences of varying lengths through a mechanism for updating internal states that considers the entire sequence history. Modern state modifications help diminish vanishing and exploding gradients.
+
+|Model                                       | Point Forecast   | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values  |
+|:-------------------------------------------|:----------------:|:----------------------:|:----------------------:|:----------------------------:|
+|[`RNN`](../models.rnn.html)                 |✅                |✅                      |✅                      |✅                            |
+|[`GRU`](../models.gru.html)                 |✅                |✅                      |✅                      |✅                            |
+|[`LSTM`](../models.lstm.html)               |✅                |✅                      |✅                      |✅                            |
+|[`TCN`](../models.tcn.html)                 |✅                |✅                      |✅                      |✅                            |
+|[`DeepAR`](../models.deepar.html)           |✅                |✅                      |✅                      |✅                            |
+|[`DilatedRNN`](../models.dilated_rnn.html)  |✅                |✅                      |✅                      |✅                            |
+: {tbl-colwidths="[25,25]"}
+
+## Transformers Model Family
+Transformer architectures are an alternative to recurrent networks. These networks build on the self-attention mechanism that directly allows modeling the relationship between different sequence parts without sequential processing. Attention makes Transformers more parallelizable than RNNs.
+
+|Model                                                      | Point Forecast   | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values  |
+|:----------------------------------------------------------|:----------------:|:----------------------:|:----------------------:|:----------------------------:|
+|[`TFT`](../models.tft.html)                                |✅                |✅                      |✅                      |✅                            |
+|[`Informer`](../models.informer.html)                      |✅                |✅                      |✅                      |✅                            |
+|[`Autoformer`](../models.autoformer.html)                  |✅                |✅                      |✅                      |✅                            |
+|[`PatchTST`](../models.patchtst.html)                      |✅                |✅                      |✅                      |✅                            |
+|[`VanillaTransformer`](../models.vanillatransformer.html)  |✅                |✅                      |✅                      |✅                            |
+: {tbl-colwidths="[25,25]"}
+
+## CNN-Based Model Family
+Convolutional Neural Networks (CNNs), originally celebrated for their accomplishments in image processing and computer vision, have also revealed substantial prowess in time series forecasting. Navigating through temporal data, CNNs utilize their convolutional layers to automatically and adaptively learn temporal patterns from the input data, offering an approach to uncovering subtle, underlying patterns embedded within a series of values.
+
+|Model                                                      | Point Forecast   | Probabilistic Forecast | Insample fitted values | Probabilistic fitted values  |
+|:----------------------------------------------------------|:----------------:|:----------------------:|:----------------------:|:----------------------------:|
+|[`TimesNet`](../models.timesnet.html)                      |✅                |✅                      |✅                      |✅                            |