The optimizer wrapper for PyTorch Adam :class:`torch.optim.Adam`. Parameters ---------- lr : float The learning rate of the optim

Base dataset class for models in PyPOTS. Parameters ---------- data : The dataset for model input, should be a dictionary or

Scaled dot-product attention. Parameters ---------- temperature: The temperature for scaling. attn_dropout: The drop

Transformer multi-head attention module. Parameters ---------- n_heads: The number of heads in multi-head attention. d_model

Series decomposition block

Generate positional encoding according to time information.

Transformer encoder layer. Parameters ---------- d_model: The dimension of the input tensor. d_ffn: The dimension of

A special designed layer normalization for the seasonal part.

The module used to generate the temporal decay factor gamma in the GRU-D model. Please refer to the original paper :cite:`che2018GRUD` for more de

Transformer encoder. Parameters ---------- n_layers: The number of layers in the encoder. d_model: The dimension of

Autoformer encoder layer with the progressive decomposition architecture.

Dataset class for BRITS. Parameters ---------- data : The dataset for model input, should be a dictionary including keys as 'X' a

Dataset class for GP-VAE. Parameters ---------- data : The dataset for model input, should be a dictionary including keys as 'X'

Dataset class for BRITS. Parameters ---------- data : The dataset for model input, should be a dictionary including keys as 'X' a

Dataset class for USGAN, the same with the one for BRITS. Parameters ---------- data : The dataset for model input, should be a d

Perform attention across the -2 dim (the -1 dim is `model_dim`). Make sure the tensor is permuted to correct shape before attention. E.g.

AutoCorrelation Mechanism with the following two phases: (1) period-based dependencies discovery (2) time delay aggregation

model BRITS: Bidirectional RITS BRITS consists of two RITS, which take time-series data from two directions (forward/backward) respectively.

model RITS: Recurrent Imputation for Time Series Attributes ---------- n_steps : sequence length (number of time steps) n_fe

Actually Autoformer uses the same data strategy as SAITS, needs MIT for training.

Dataset for CSDI model. Notes ----- In CSDI official code, `observed_mask` indicates all observed values in raw data. `gt_mask` indic

Actually Crossformer uses the same data strategy as SAITS, needs MIT for training.

Actually DLinear uses the same data strategy as SAITS, needs MIT for training.

Actually ETSformer uses the same data strategy as SAITS, needs MIT for training.

Actually FEDformer uses the same data strategy as SAITS, needs MIT for training.

Actually Informer uses the same data strategy as SAITS, needs MIT for training.

Actually PatchTST uses the same data strategy as SAITS, needs MIT for training.

Dataset for models that need MIT (masked imputation task) in their training, such as SAITS. For more information about MIT, please refer to :cite

Actually TimesNet uses the same data strategy as SAITS, needs MIT for training.

1D multiwavelet block.

Notes ----- This implementation is adapted from https://gist.github.com/Kaixhin/57901e91e5c5a8bac3eb0cbbdd3aba81

Position-wise feed forward network (FFN) in Transformer. Parameters ---------- d_in: The dimension of the input tensor. d_hi

The original positional-encoding module for Transformer. Parameters ---------- d_hid: The dimension of the hidden layer. n_p

PyTorch Dataset for fitting timeseries models. The dataset automates common tasks such as * scaling and encoding of variables * nor

Autoformer decoder layer with the progressive decomposition architecture

model GPVAE with Gaussian Process prior Parameters ---------- input_dim : int, the feature dimension of the input time_lengt

USGAN model

The PyTorch implementation of the CSDI model :cite:`tashiro2021csdi`. Parameters ---------- n_steps : The number of time steps in

The PyTorch implementation of the DLinear model. DLinear is originally proposed by Zeng et al. in :cite:`zeng2023dlinear`. Parameters ---

A simple fully-connected layer.

Spatial embedded attention layer. The node embedding serves as the query and key matrices for attentive aggregation on graphs.

The module used to capture the correlation between features for imputation in BRITS. Attributes ---------- W : tensor The weights

The PyTorch implementation of the GPVAE model :cite:`fortuin2020gpvae`. Parameters ---------- n_steps : The number of time steps

The PyTorch implementation of the ImputeFormer model. ImputeFormer is originally proposed by Nie et al. in KDD'24: cite:`nie2024imputeformer`.

Sets the learning rate of each parameter group to the initial lr times a given function. When last_epoch=-1, sets initial lr as lr. Parameter

Simple Multi-layer Perceptron encoder with optional linear readout.

The moving average block to highlight the trend of time series.