timeseries

Timeseries datasets.

Timeseries data is special and has to be processed and fed to algorithms in a special way. This module defines a class that is able to handle a wide variety of timeseries data problems.

Functions

_find_end_indices(diffs, max_lengths, min_length)	Identify end indices in series even if some values are missing.
check_for_nonfinite(tensor, names)	Check if 2D tensor contains NAs or inifinite values.

Classes

TimeSeriesDataSet(data, time_idx, target, …)	PyTorch Dataset for fitting timeseries models.
TimeSynchronizedBatchSampler(data_source, …)	Samples mini-batches randomly but in a time-synchronised manner.

pytorch_forecasting.data.examples.get_stallion_data pytorch_forecasting.data.timeseries._find_end_indices