Machine Learning Algorithms - Second Edition

Precision

Another useful direct measure is the precision (or positive predictive value):

This is directly connected to the ability to capture features that determine the positiveness of a sample, to avoid misclassification as negative. In scikit-learn, the implementation is based on the previous binary example:

from sklearn.metrics import precision_scoreprint(precision_score(Y_test, lr.predict(X_test)))0.953

If you don't flip the confusion matrix, but want to get the same measures, it's necessary to add the pos_label=0 parameter to all metric score functions.