Featuretools is best for relational categorical and numerical data.
Regarding text it seems that it only counts text length and some other very basic stats.
What would be the best pipeline for preparing textual data for featuretools?
Should it be done with make_trans_primitive or it would be better to prepare data some other way?
