Specificity: true negative rate

Beginner Netflix, Spotify+5

The problem, from first principles

Recall answers "of the real positives, how many did we find?". Specificity answers the mirror question for the negative class: "of the real negatives, how many did we correctly leave alone?". A spam filter with high recall and low specificity catches spam but buries legitimate mail.

Implement specificity(y_true, y_pred, positive=1). Treat positive as the positive class and everything else as negative. Return TN / (TN + FP).

Constraints

If there are no negative rows, return 0.0.
Inputs must have the same length, otherwise raise ValueError.

Hints

Open one at a time. Each gives away a little more than the last.

Hint 1

Negative rows are y_true != positive. Among them, a prediction counts as a true negative when it is also not positive.

Hint 2

np.mean(y_pred[neg] != positive) over the negative rows gives the same ratio directly.

Keywords

classical-mlevaluationclassification

Python

Run your code to see results

No results yet

Run your code to check it against the examples, or submit to run every test.