torch_timeseries.dataloader.SlidingWindowTS

class SlidingWindowTS(dataset: TimeSeriesDataset, scaler: Scaler, time_enc=1, window: int = 168, horizon: int = 1, steps: int = 2, scale_in_train=True, shuffle_train=True, freq=None, batch_size: int = 32, train_ratio: float = 0.7, test_ratio: float = 0.2, val_ratio: float | None = None, num_worker: int = 3, uniform_eval=True, single_variate=False, fast_train=False, fast_val=False, fast_test=False, include_raw=True, time_index=False)[source]

Bases: object

Class for splitting the dataset sequentially and then randomly sampling from each subset.

time_index: with two addtional values of time index, (…, x_index, y_index)

dataset

Time series dataset to be used.

Type:: TimeSeriesDataset

scaler

Scaler to normalize the data.

Type:: Scaler

time_enc

Time encoding flag.

Type:: int

window

Window size for the time series data.

Type:: int

horizon

Forecast horizon.

Type:: int

steps

Step size between windows.

Type:: int

scale_in_train

Whether to scale data during training.

Type:: bool

shuffle_train

Whether to shuffle the training data.

Type:: bool

freq

Frequency of the time series data.

Type:: str or None

batch_size

Number of samples per batch.

Type:: int

train_ratio

Ratio of the dataset to be used for training.

Type:: float

val_ratio

Ratio of the dataset to be used for validation.

Type:: float

test_ratio

Ratio of the dataset to be used for testing.

Type:: float

num_worker

Number of worker threads for data loading.

Type:: int

uniform_eval

Whether to use uniform evaluation.

Type:: bool

train_loader

DataLoader for the training data.

Type:: DataLoader

val_loader

DataLoader for the validation data.

Type:: DataLoader

test_loader

DataLoader for the test data.

Type:: DataLoader