CODEWITHSUNDEEP
pythonscikit-learncross-validation
user1566200
user1566200

Reputation: 1838

Cross_val_predict in SKLearn - expected an array but

I have a very simple pandas DataFrame (X), with eight columns, and twenty rows full of floats. Then, I have another DataFrame (y), that is twenty rows. When I run:

score = cross_val_score(clf, X, y, scoring='accuracy')

I get no error and the results make sense (there is nothing wrong with the DataFrame, basically). However, when I run:

predicted = cross_val_predict(clf, X)

or

predicted = cross_val_predict(clf, X.values)

I get the following error:

Expected array-like (array or non-string sequence), got None

Any suggestions as to what the problem might be?

Upvotes: 0

Views: 855

Answers (1)

Christian Hirsch
Christian Hirsch

Reputation: 2056

Since you are dealing with a supervised learning problem, you should also provide y as an argument to cross_val_predict, i.e., cross_val_predict(clf, X,y). This is necessary, as k fold cross-validation requires training the classifier for multiple training sets.

Upvotes: 2

Related Questions