Cumulative Gains

Problem

Compute the cumulative gains curve: sort by score descending (ties by original index) and return the cumulative share of positives captured after each sample.

Return [] when there are no positive labels or the input is empty.

Hint

Gains rise fastest when the highest-scored samples are the true positives.

Test cases

case 1

in: [[0,0,1,1],[0.1,0.4,0.8,0.9]]

expected: [0.5,1,1,1]

case 2

in: [[1,0,1],[0.5,0.9,0.5]]

expected: [0,0.5,1]

case 3

in: [[0,0],[0.1,0.2]]

expected: []

case 4

in: [[],[]]

expected: []

Starter codePython

def cumulative_gains(y_true, scores):
    # Your code here
    pass

4 test cases · run your code against them in the editor