An error is thrown when using Random forest classifier multi-target non-numeric classes #11451
Description
Description
Random forest classifier can be fit on multiple target columns where columns have string labels and are not numeric, but when trying to predict, an error is thrown.
Steps/Code to Reproduce
import numpy as np
from sklearn.datasets import make_classification
from sklearn.ensemble.forest import RandomForestClassifier
c = RandomForestClassifier()
X, y = make_classification()
y = np.array(['foo' if v else 'bar' for v in y]).reshape((y.shape[0], 1))
ys = np.hstack([y, y])
c.fit(X, ys)
c.predict(X)Expected Results
No error is thrown.
Actual Results
File ".../local/lib/python2.7/site-packages/sklearn/ensemble/forest.py", line 550, in predict
axis=0)
ValueError: could not convert string to float: foo
Versions
Linux-4.13.0-45-generic-x86_64-with-Ubuntu-17.10-artful
('Python', '2.7.14 (default, Sep 23 2017, 22:06:14) \n[GCC 7.2.0]')
('NumPy', '1.14.5')
('SciPy', '1.1.0')
('Scikit-Learn', '0.19.1')