This is controlled by self.default_features and self.custom_features.
However, because of complexity (e.g., some features, like information exchange, depend on lexical features being calculated first), we have decided to leave the ability to choose features you do NOT want for a later issue.
This feature thus encapsulates having an additional parameter called self.excluded_features() that would exclude features from being calculated if the user doesn't want them. Currently, there is some preprocessing (lexicons, BERT features) that is always done, because the relevant features are in the default set. If the user wants to drop unwanted features, we could potentially make the FeatureBuilder run faster.
Per https://github.com/Watts-Lab/team-process-map/issues/202, we implemented the ability for users to specify additional features they DO want that are not in the default set.
This is controlled by
self.default_features
andself.custom_features
.However, because of complexity (e.g., some features, like information exchange, depend on lexical features being calculated first), we have decided to leave the ability to choose features you do NOT want for a later issue.
This feature thus encapsulates having an additional parameter called
self.excluded_features()
that would exclude features from being calculated if the user doesn't want them. Currently, there is some preprocessing (lexicons, BERT features) that is always done, because the relevant features are in the default set. If the user wants to drop unwanted features, we could potentially make the FeatureBuilder run faster.