Grouped CV Fold Sizes

Problem

Compute the sample count of every fold in a grouped k-fold split. Sort the distinct groups and assign them to contiguous folds so the first (g % k) folds receive one extra group, where g is the number of groups. A fold's size is the total number of samples in its groups. Return [] when k <= 0.

Hint

Keeping every group inside one fold prevents the same entity leaking across train and validation.

Test cases

case 1

in: [["a","a","b","b","c","c"],2]

expected: [4,2]

case 2

in: [["a","a","a","b","c"],2]

expected: [4,1]

case 3

in: [["x","y"],5]

expected: [1,1,0,0,0]

case 4

in: [[],2]

expected: [0,0]

case 5

in: [["m","m"],0]

expected: []

Starter codePython

def grouped_fold_sizes(groups, k):
    # Your code here
    pass

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