precision and recall error while using sklearn

Question

I am using sklearn precision and recall to get those scores. I got an error saying value error. Could anyone tell where am I doing wrong?

My y_test is as follows

443     positive
3615    positive
2030    negative
2993    positive
2870    positive
2907    negative
2215    positive

My Prediction is as follows

['positive' 'positive' 'positive' ..., 'positive' 'positive' 'positive']

Code:

from sklearn.metrics import precision_score
from sklearn.metrics import recall_score
precision_score(y_test, pred)

Error:

ValueError: pos_label=None is not a valid label: array(['negative', 'positive'], 
      dtype='<U8')

Vivek Kumar Vivek Kumar · Accepted Answer · 2018-05-24T11:51:17

Precision is defined as ratio of true positives to total predicted positives.

precision = tp / (tp + fp)

Now in your case, the program dont know which label is to be considered as positive class. So you need to define it yourself. Do this:

precision_score(y_test, pred, pos_label='positive')

Also, the error you shown: pos_label=None is not a valid label tells that you may have a older version of scikit. Newer versions should throw this error (if pos_label not specified):

pos_label=1 is not a valid label

So I would advice you to upgrade to latest version

precision and recall error while using sklearn

1 Answers