parsnip.load_dataset

parsnip.load_dataset(path, kind=None, in_memory=True, reject_invalid=True, require_redshift=True, label_map=None, valid_classes=None, verbose=True)

Load a dataset using the lcdata package.

This can be any lcdata HDF5 dataset. We use parse_dataset to clean things up for ParSNIP by rejecting irrelevant light curves (e.g. galactic ones) and updating class labels.

We try to guess the dataset type from the filename. If this doesn’t work, specify the filename explicitly instead.

Parameters:

  • path (str) – Path to the dataset on disk

  • kind (str, optional) – Kind of dataset, by default we will attempt to determine it from the filename

  • in_memory (bool, optional) – If False, don’t load the light curves into memory, and only load the metadata. See lcdata.Dataset for details.

  • reject_invalid (bool, optional) – Whether to reject invalid light curves, by default True

  • label_map (dict, optional) – Overwriting the default classification label mapping with a custom dict

  • verbose (bool, optional) – If True, print parsing information, by default True

Returns:

Loaded dataset

Return type:

Dataset