Train-Serve Skew Ratio

Problem

Compute the worst relative train-serve feature skew: the maximum over features of abs(serve_mean - train_mean) / max(abs(train_mean), 1e-9). Compare only up to the shorter input; empty input returns 0.0.

Hint

Feature statistics that differ between training and serving pipelines cause silent degradation.

Test cases

case 1

in: [[1,2,3],[1.1,1.9,3.3]]

expected: 0.10000000000000009

case 2

in: [[0,5],[0,10]]

expected: 1

case 3

in: [[2],[2]]

expected: 0

case 4

in: [[],[]]

expected: 0

Starter codePython

def train_serve_skew(train_means, serve_means):
    # Your code here
    pass

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