问题
I'm currently using sklearn library in python to use supervised machine learning. I have a list of records like this:
[x1, x2, x3] -> [y1]
And i'm using the Bag Of Words technique. It all works. Sometimes it could happen that the user says the prediction is not right.
Something like a negative record:
[x1, x2, x3] != [y1]
I would like that if this happens the next time (or after many negative feedbacks) the same prediction won't appear.
来源：https://stackoverflow.com/questions/45545178/make-better-machine-learning-prediction-thanks-to-negative-feedback