Walk-Forward Fold Count

Problem

Count how many non-overlapping walk-forward folds fit in a series of n observations, given an initial training size and a test horizon h. Each fold trains on the observations available so far and tests the next h points, so the count is (n - initial) // h.

Return 0 when h is not positive, when n is not positive, or when the initial size is negative or exceeds n.

Hint

Walk-forward evaluation uses every point exactly once as a test target.

Test cases

case 1

in: [10,4,2]

expected: 3

case 2

in: [10,5,1]

expected: 5

case 3

in: [5,5,2]

expected: 0

case 4

in: [10,4,0]

expected: 0

Starter codePython

def walk_forward_fold_count(n, initial, h):
    # Your code here
    pass

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