How to use TPOT in the text domain?

[ben0it8]

Hello,

My question is if it's possible to use TPOT in the text domain for classification task?
Given a labeled corpus (eg. label - document pairs) I'd like to perform classification to infer the label of an unseen piece of document.

Thanks,
Oliver

[weixuanfu]

I think this issue is related to #507. We are working on a configurable grammar in #523 to add the support for text classification. For now, you may try to transform text to numeric matrix using CountVectorizer, TFIDFVectorizer and HashingVectorizer before using TPOTClassifier for your problem.