window_generator#

Module that generates specifiable windows from time-series data.

create_array_window(data: ndarray[Any, dtype[float64]], input_length: int, output_length: int) → ndarray[Any, dtype[float64]]#

Creates a 3 dimensional array of windows out of a single time series.

Parameters:

data – Time series of the shape (n,m), with n observations and m attributes.
input_length – Length of the input window.
output_length – Length of the output window.

Returns:

Time series transformed into an 3 dimensional windowing array, (x,y,z) with x: windows, y: time points, z: attributes

Return type:

window

create_window_dataset(data_list: list[numpy.ndarray[Any, numpy.dtype[numpy.float64]]], input_length: int, output_length: int) → tuple[numpy.ndarray[Any, numpy.dtype[numpy.float64]], numpy.ndarray[Any, numpy.dtype[numpy.float64]]]#

Creates dataset in the form of a 3 dimensional array of windows.

Parameters:

data_list – 3D array of time series in the shape (x,y,z), with x observations y timestamps and m attributes.
input_length – Length of the input window.
output_length – Length of the output window.

Returns:

Input data of the shape (x,y,z),: with x: windows, y: time points, z: attributes
y: Output data of the shape (x,y,z),: with x: windows, y: time points, z: attributes

Return type: