machine learning - Custom Features using scikit-learn -
i working on project classify short text. 1 requirement have along vectorizing short text, add additional feature length of text, number of url's etc features each input.
is supported in scikit-learn? link example notebook or video help.
thanks, romit.
you can combine features extracted different transfomers (e.g. 1 extracts bag of words (bow) features 1 extracts other statistics) using featureunion class.
the normalization of features , there small number respect number of distinct bow features problematic. whether or not problem depends on assumptions made models trained downstream , on specific data , target task.
Comments
Post a Comment