avocado.Featurizer.extract_raw_features

Featurizer.extract_raw_features(astronomical_object, return_model=False)

Extract raw features from an object

Featurizing is slow, so the idea here is to extract a lot of different things, and then in select_features these features are postprocessed to select the ones that are actually fed into the classifier. This allows for rapid iteration of training on different feature sets. Note that the features produced by this method are often unsuitable for classification, and may include data leaks. Make sure that you understand what features can be used for real classification before making any changes.

For now, there is no generic featurizer, so this must be implemented in survey-specific subclasses.

Parameters:

• astronomical_object (AstronomicalObject) – The astronomical object to featurize.
• return_model (bool) – If true, the light curve model is also returned. Defaults to False.

Returns:

• raw_features (dict) – The raw extracted features for this object.
• model (dict (optional)) – A dictionary with the light curve model in each band. This is only returned if return_model is set to True.