SOFTSSharp

SOFTSSharp: SOFTS extension with stochastic variable-position encoding for multivariate time series forecasting.

SOFTSSharp extends SOFTS by stochastically adding variable-position embeddings and multiple dropout layers inside the STAD aggregation-redistribution component, aiming to improve forecasting accuracy while preserving linear complexity.

Figure 1. Architecture of SOFTSSharp

1. SOFTSSharp

SOFTSSharp

SOFTSSharp(
    h,
    input_size,
    n_series,
    futr_exog_list=None,
    hist_exog_list=None,
    stat_exog_list=None,
    exclude_insample_y=False,
    hidden_size=512,
    d_core=512,
    e_layers=2,
    d_ff=2048,
    dropout=0.1,
    pe_keep_prob=0.5,
    use_norm=True,
    loss=MAE(),
    valid_loss=None,
    max_steps=1000,
    learning_rate=0.001,
    num_lr_decays=-1,
    early_stop_patience_steps=-1,
    val_monitor="ptl/val_loss",
    val_check_steps=100,
    batch_size=32,
    valid_batch_size=None,
    windows_batch_size=32,
    inference_windows_batch_size=32,
    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

SOFTSSharp

SOFTS# (SOFTSSharp) extends SOFTS by stochastically adding variable-position embeddings and multiple dropout layers inside the STAD component.

Parameters:

Name	Type	Description	Default
h	int	Forecast horizon.	required
input_size	int	Autoregressive inputs size.	required
n_series	int	Number of time-series.	required
hidden_size	int	Dimension of the model.	512
d_core	int	Dimension of core in STADSharp.	512
e_layers	int	Number of encoder layers.	2
d_ff	int	Dimension of fully-connected layer.	2048
dropout	float	Dropout rate.	0.1
pe_keep_prob	float	probability of applying variable-position encoding during training. During inference, the positional encoding is scaled by this value.	0.5
use_norm	bool	Whether to normalize or not.	True
loss	PyTorch module	Instantiated train loss class from losses collection.	MAE()
valid_loss	PyTorch module	Instantiated valid loss class from losses collection.	None
max_steps	int	Maximum number of training steps.	1000
learning_rate	float	Learning rate between (0, 1).	0.001
num_lr_decays	int	Number of learning rate decays, evenly distributed across max_steps.	-1
early_stop_patience_steps	int	Number of validation iterations before early stopping.	-1
val_monitor	str	Metric to monitor for early stopping.	‘ptl/val_loss’
val_check_steps	int	Number of training steps between every validation loss check.	100
batch_size	int	Number of different series in each batch.	32
valid_batch_size	int	Number of different series in each validation and test batch, if None uses batch_size.	None
windows_batch_size	int	Number of windows to sample in each training batch, default uses all.	32
inference_windows_batch_size	int	Number of windows to sample in each inference batch, -1 uses all.	32
start_padding_enabled	bool	If True, the model will pad the time series with zeros at the beginning, by input size.	False
step_size	int	Step size between each window of temporal data.	1
scaler_type	str	Type of scaler for temporal inputs normalization.	‘identity’
random_seed	int	Random seed for pytorch initializer and numpy generators.	1
drop_last_loader	bool	If True TimeSeriesDataLoader drops last non-full batch.	False
alias	str	Optional custom name of the model.	None

SOFTSSharp.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:

Name	Type	Description	Default
dataset	TimeSeriesDataset	NeuralForecast’s TimeSeriesDataset, see documentation.	required
val_size	int	Validation size for temporal cross-validation.	0
random_seed	int	Random seed for pytorch initializer and numpy generators, overwrites model.init’s.	None
test_size	int	Test size for temporal cross-validation.	0

Returns:

Type	Description
None

SOFTSSharp.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:

Name	Type	Description	Default
dataset	TimeSeriesDataset	NeuralForecast’s TimeSeriesDataset, see documentation.	required
test_size	int	Test size for temporal cross-validation.	None
step_size	int	Step size between each window.	1
random_seed	int	Random seed for pytorch initializer and numpy generators, overwrites model.init’s.	None
quantiles	list	Target quantiles to predict.	None
h	int	Prediction horizon, if None, uses the model’s fitted horizon. Defaults to None.	None
explainer_config	dict	configuration for explanations.	None
**data_module_kwargs	dict	PL’s TimeSeriesDataModule args, see documentation.

Returns:

Type	Description
None

Usage example

import pandas as pd
import matplotlib.pyplot as plt

from neuralforecast import NeuralForecast
from neuralforecast.models import SOFTSSharp
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic
from neuralforecast.losses.pytorch import MASE

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

model = SOFTSSharp(h=12,
                   input_size=24,
                   n_series=2,
                   hidden_size=256,
                   d_core=256,
                   e_layers=2,
                   d_ff=64,
                   dropout=0.1,
                   pe_keep_prob=0.5,
                   use_norm=True,
                   loss=MASE(seasonality=4),
                   early_stop_patience_steps=3,
                   batch_size=32)

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

fig, ax = plt.subplots(1, 1, figsize = (20, 7))
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])

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['SOFTSSharp'], c='blue', label='Forecast')
ax.set_title('AirPassengers Forecast', fontsize=22)
ax.set_ylabel('Monthly Passengers', fontsize=20)
ax.set_xlabel('Year', fontsize=20)
ax.legend(prop={'size': 15})
ax.grid()

2. Auxiliary functions

PositionalEmbedding

PositionalEmbedding(d_series, max_len=5000)

Bases: Module

STADSharp

STADSharp(d_series, d_core, dropout_rate=0.1, pe_keep_prob=0.5)

Bases: Module

STar Aggregate Dispatch Module with stochastic variable-position encoding.