dqa.tasks.time_series.WindowSplit#

class dqa.tasks.time_series.WindowSplit(window_length: int, overlap: int = 0, keep_size: bool = False, sampling_time_name: Optional[str] = None, **kwargs)#

Splits input arrays into windows of a fixed length.

For an input array of shape (length, …), the result has shape (number_of_windows, …, window_length). The windows generated in this process can be chosen to have a specified overlap.

Parameters:
  • window_length (int) – The length of each of the windows.

  • overlap (int, default=0) – The number of indices for the overlap of the windows.

  • keep_size (bool, default=False) – If set, the size along the first axis remains the same. In this case, the overlap is set to window_length - 1. The array is padded with zero entries along axis 0 to match the required size.

  • sampling_time_name (str, optional, default=None) – Name of the sampling time field in the metadata. If set, then this field will be adjusted according to the rearrangement of the data.

Methods

finish()

Can perform actions that are required to clean up after the task has finished, e.g.

in_out_default

input_output_dataset

input_output_machine

input_output_mode

input_output_name

log

modify_data_row

modify_dataset

modify_dataset_dict

modify_machine

modify_measurement

set_logging_level

set_new_metadata_sampling_time

transfer_metadata