Machine Learning Algorithms - Second Edition

Recall

The ability to detect true positive samples among all the potential positives (including the false negatives) can be assessed using another measure called recall (sometimes defined as sensitivity):

The scikit-learn implementation is as follows:

from sklearn.metrics import recall_scoreprint(recall_score(Y_test, lr.predict(X_test)))0.984

It's not surprising that we have a 98% recall with 95% precision because the number of false negatives (which directly impact the recall) is proportionally lower (1) than the number of false positives (3), which impact the precision.

The reader should understand the subtle difference between these two ...