Skip to main content
The PatchTST model is an efficient Transformer-based model for multivariate time series forecasting. It is based on two key components: - segmentation of time series into windows (patches) which are served as input tokens to Transformer - channel-independence. where each channel contains a single univariate time series. References Figure 1. PatchTST. Figure 1. PatchTST.

​
1. PatchTST

​
PatchTST

PatchTST(
    h,
    input_size,
    stat_exog_list=None,
    hist_exog_list=None,
    futr_exog_list=None,
    exclude_insample_y=False,
    encoder_layers=3,
    n_heads=16,
    hidden_size=128,
    linear_hidden_size=256,
    dropout=0.2,
    fc_dropout=0.2,
    head_dropout=0.0,
    attn_dropout=0.0,
    patch_len=16,
    stride=8,
    revin=True,
    revin_affine=False,
    revin_subtract_last=True,
    activation="gelu",
    res_attention=True,
    batch_normalization=False,
    learn_pos_embed=True,
    loss=MAE(),
    valid_loss=None,
    max_steps=5000,
    learning_rate=0.0001,
    num_lr_decays=-1,
    early_stop_patience_steps=-1,
    val_check_steps=100,
    batch_size=32,
    valid_batch_size=None,
    windows_batch_size=1024,
    inference_windows_batch_size=1024,
    start_padding_enabled=False,
    training_data_availability_threshold=0.0,
    step_size=1,
    scaler_type="identity",
    random_seed=1,
    drop_last_loader=False,
    alias=None,
    optimizer=None,
    optimizer_kwargs=None,
    lr_scheduler=None,
    lr_scheduler_kwargs=None,
    dataloader_kwargs=None,
    **trainer_kwargs
)
Bases: BaseModel PatchTST The PatchTST model is an efficient Transformer-based model for multivariate time series forecasting. It is based on two key components:
  • segmentation of time series into windows (patches) which are served as input tokens to Transformer
  • channel-independence, where each channel contains a single univariate time series.
Parameters:
NameTypeDescriptionDefault
hintforecast horizon.required
input_sizeintautorregresive inputs size, y=[1,2,3,4] input_size=2 -> y_[t-2:t]=[1,2].required
stat_exog_liststr liststatic exogenous columns.None
hist_exog_liststr listhistoric exogenous columns.None
futr_exog_liststr listfuture exogenous columns.None
exclude_insample_yboolthe model skips the autoregressive features y[t-input_size:t] if True.False
encoder_layersintnumber of layers for encoder.3
n_headsintnumber of multi-head’s attention.16
hidden_sizeintunits of embeddings and encoders.128
linear_hidden_sizeintunits of linear layer.256
dropoutfloatdropout rate for residual connection.0.2
fc_dropoutfloatdropout rate for linear layer.0.2
head_dropoutfloatdropout rate for Flatten head layer.0.0
attn_dropoutfloatdropout rate for attention layer.0.0
patch_lenintlength of patch. Note: patch_len = min(patch_len, input_size + stride).16
strideintstride of patch.8
revinboolbool to use RevIn.True
revin_affineboolbool to use affine in RevIn.False
revin_subtract_lastboolbool to use substract last in RevIn.True
activationstractivation from [‘gelu’,‘relu’].‘gelu’
res_attentionboolbool to use residual attention.True
batch_normalizationboolbool to use batch normalization.False
learn_pos_embedboolbool to learn positional embedding.True
lossPyTorch moduleinstantiated train loss class from losses collection.MAE()
valid_lossPyTorch moduleinstantiated valid loss class from losses collection.None
max_stepsintmaximum number of training steps.5000
learning_ratefloatlearning rate between (0, 1).0.0001
num_lr_decaysintnumber of learning rate decays, evenly distributed across max_steps.-1
early_stop_patience_stepsintnumber of validation iterations before early stopping.-1
val_check_stepsintnumber of training steps between every validation loss check.100
batch_sizeintnumber of different series in each batch.32
valid_batch_sizeintnumber of different series in each validation and test batch, if None uses batch_size.None
windows_batch_sizeintnumber of windows to sample in each training batch, default uses all.1024
inference_windows_batch_sizeintnumber of windows to sample in each inference batch.1024
start_padding_enabledboolif True, the model will pad the time series with zeros at the beginning, by input size.False
training_data_availability_thresholdUnion[float, List[float]]minimum fraction of valid data points required for training windows. Single float applies to both insample and outsample; list of two floats specifies [insample_fraction, outsample_fraction]. Default 0.0 allows windows with only 1 valid data point (current behavior).0.0
step_sizeintstep size between each window of temporal data.1
scaler_typestrtype of scaler for temporal inputs normalization see temporal scalers.‘identity’
random_seedintrandom_seed for pytorch initializer and numpy generators.1
drop_last_loaderboolif True TimeSeriesDataLoader drops last non-full batch.False
aliasstroptional, Custom name of the model.None
optimizerSubclass of ‘torch.optim.Optimizer’optional, user specified optimizer instead of the default choice (Adam).None
optimizer_kwargsdictoptional, list of parameters used by the user specified optimizer.None
lr_schedulerSubclass of ‘torch.optim.lr_scheduler.LRScheduler’optional, user specified lr_scheduler instead of the default choice (StepLR).None
lr_scheduler_kwargsdictoptional, list of parameters used by the user specified lr_scheduler.None
dataloader_kwargsdictoptional, list of parameters passed into the PyTorch Lightning dataloader by the TimeSeriesDataLoader.None
**trainer_kwargsintkeyword trainer arguments inherited from PyTorch Lighning’s trainer.

​
PatchTST.fit

fit(
    dataset, val_size=0, test_size=0, random_seed=None, distributed_config=None
)
Fit. The fit method, optimizes the neural network’s weights using the initialization parameters (learning_rate, windows_batch_size, …) and the loss function as defined during the initialization. Within fit we use a PyTorch Lightning Trainer that inherits the initialization’s self.trainer_kwargs, to customize its inputs, see PL’s trainer arguments. The method is designed to be compatible with SKLearn-like classes and in particular to be compatible with the StatsForecast library. By default the model is not saving training checkpoints to protect disk memory, to get them change enable_checkpointing=True in __init__. Parameters:
NameTypeDescriptionDefault
datasetTimeSeriesDatasetNeuralForecast’s TimeSeriesDataset, see documentation.required
val_sizeintValidation size for temporal cross-validation.0
random_seedintRandom seed for pytorch initializer and numpy generators, overwrites model.init’s.None
test_sizeintTest size for temporal cross-validation.0
Returns:
TypeDescription
None

​
PatchTST.predict

predict(
    dataset,
    test_size=None,
    step_size=1,
    random_seed=None,
    quantiles=None,
    h=None,
    explainer_config=None,
    **data_module_kwargs
)
Predict. Neural network prediction with PL’s Trainer execution of predict_step. Parameters:
NameTypeDescriptionDefault
datasetTimeSeriesDatasetNeuralForecast’s TimeSeriesDataset, see documentation.required
test_sizeintTest size for temporal cross-validation.None
step_sizeintStep size between each window.1
random_seedintRandom seed for pytorch initializer and numpy generators, overwrites model.init’s.None
quantileslistTarget quantiles to predict.None
hintPrediction horizon, if None, uses the model’s fitted horizon. Defaults to None.None
explainer_configdictconfiguration for explanations.None
**data_module_kwargsdictPL’s TimeSeriesDataModule args, see documentation.
Returns:
TypeDescription
None

​
Usage example

import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import PatchTST
from neuralforecast.losses.pytorch import DistributionLoss
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic, augment_calendar_df

AirPassengersPanel, calendar_cols = augment_calendar_df(df=AirPassengersPanel, freq='M')

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = PatchTST(h=12,
                 input_size=104,
                 patch_len=24,
                 stride=24,
                 revin=False,
                 hidden_size=16,
                 n_heads=4,
                 scaler_type='robust',
                 loss=DistributionLoss(distribution='StudentT', level=[80, 90]),
                 learning_rate=1e-3,
                 max_steps=500,
                 val_check_steps=50,
                 early_stop_patience_steps=2)

nf = NeuralForecast(
    models=[model],
    freq='ME'
)
nf.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)
forecasts = nf.predict(futr_df=Y_test_df)

Y_hat_df = forecasts.reset_index(drop=False).drop(columns=['unique_id','ds'])
plot_df = pd.concat([Y_test_df, Y_hat_df], axis=1)
plot_df = pd.concat([Y_train_df, plot_df])

if model.loss.is_distribution_output:
    plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
    plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
    plt.plot(plot_df['ds'], plot_df['PatchTST-median'], c='blue', label='median')
    plt.fill_between(x=plot_df['ds'][-12:], 
                    y1=plot_df['PatchTST-lo-90'][-12:].values, 
                    y2=plot_df['PatchTST-hi-90'][-12:].values,
                    alpha=0.4, label='level 90')
    plt.grid()
    plt.legend()
    plt.plot()
else:
    plot_df = plot_df[plot_df.unique_id=='Airline1'].drop('unique_id', axis=1)
    plt.plot(plot_df['ds'], plot_df['y'], c='black', label='True')
    plt.plot(plot_df['ds'], plot_df['PatchTST'], c='blue', label='Forecast')
    plt.legend()
    plt.grid()

​
2. Backbone

​
Auxiliary Functions

​
get_activation_fn

get_activation_fn(activation)

​
Transpose

Transpose(*dims, contiguous=False)
Bases: Module Transpose

​
Positional Encoding

​
positional_encoding

positional_encoding(pe, learn_pe, q_len, hidden_size)

​
Coord1dPosEncoding

Coord1dPosEncoding(q_len, exponential=False, normalize=True)

​
Coord2dPosEncoding

Coord2dPosEncoding(
    q_len, hidden_size, exponential=False, normalize=True, eps=0.001
)

​
PositionalEncoding

PositionalEncoding(q_len, hidden_size, normalize=True)

​
Encoder

​
TSTEncoderLayer

TSTEncoderLayer(
    q_len,
    hidden_size,
    n_heads,
    d_k=None,
    d_v=None,
    linear_hidden_size=256,
    store_attn=False,
    norm="BatchNorm",
    attn_dropout=0,
    dropout=0.0,
    bias=True,
    activation="gelu",
    res_attention=False,
    pre_norm=False,
)
Bases: Module TSTEncoderLayer

​
TSTEncoder

TSTEncoder(
    q_len,
    hidden_size,
    n_heads,
    d_k=None,
    d_v=None,
    linear_hidden_size=None,
    norm="BatchNorm",
    attn_dropout=0.0,
    dropout=0.0,
    activation="gelu",
    res_attention=False,
    n_layers=1,
    pre_norm=False,
    store_attn=False,
)
Bases: Module TSTEncoder

​
TSTiEncoder

TSTiEncoder(
    c_in,
    patch_num,
    patch_len,
    max_seq_len=1024,
    n_layers=3,
    hidden_size=128,
    n_heads=16,
    d_k=None,
    d_v=None,
    linear_hidden_size=256,
    norm="BatchNorm",
    attn_dropout=0.0,
    dropout=0.0,
    act="gelu",
    store_attn=False,
    key_padding_mask="auto",
    padding_var=None,
    attn_mask=None,
    res_attention=True,
    pre_norm=False,
    pe="zeros",
    learn_pe=True,
)
Bases: Module TSTiEncoder

​
Flatten_Head

Flatten_Head(individual, n_vars, nf, h, c_out, head_dropout=0)
Bases: Module Flatten_Head

​
PatchTST_backbone

PatchTST_backbone(
    c_in,
    c_out,
    input_size,
    h,
    patch_len,
    stride,
    max_seq_len=1024,
    n_layers=3,
    hidden_size=128,
    n_heads=16,
    d_k=None,
    d_v=None,
    linear_hidden_size=256,
    norm="BatchNorm",
    attn_dropout=0.0,
    dropout=0.0,
    act="gelu",
    key_padding_mask="auto",
    padding_var=None,
    attn_mask=None,
    res_attention=True,
    pre_norm=False,
    store_attn=False,
    pe="zeros",
    learn_pe=True,
    fc_dropout=0.0,
    head_dropout=0,
    padding_patch=None,
    pretrain_head=False,
    head_type="flatten",
    individual=False,
    revin=True,
    affine=True,
    subtract_last=False,
)
Bases: Module PatchTST_backbone