It would likely benefit models like those in https://github.com/related-sciences/nxontology-ml/pull/28 if they included the presence of EFO TA terms as features. Specifically, a single binary indicator for all TA terms associated with a single EFO term would be great (i.e. following the semantics of MultiLabelBinarizer).
refs: https://github.com/related-sciences/nxontology-ml/issues/2
It would likely benefit models like those in https://github.com/related-sciences/nxontology-ml/pull/28 if they included the presence of EFO TA terms as features. Specifically, a single binary indicator for all TA terms associated with a single EFO term would be great (i.e. following the semantics of MultiLabelBinarizer).