PIT Histogram Uniformity

Problem

Check the uniformity of probability integral transform values with a chi-square-style statistic: divide [0, 1) into equal bins, count values per bin, and return sum((count - expected)^2 / expected) with expected = n / bins.

Return 0.0 for empty input or bins <= 0. Values exactly 1.0 fall into the last bin.

Hint

Perfectly calibrated forecasts produce a flat PIT histogram.

Test cases

case 1

in: [[0.1,0.2,0.3,0.4],4]

expected: 4

case 2

in: [[0.1,0.1,0.1,0.1],4]

expected: 12

case 3

in: [[0.5],2]

expected: 1

case 4

in: [[],4]

expected: 0

Starter codePython

def pit_histogram_uniformity(pit_values, bins):
    # Your code here
    pass

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