MediumTime Seriests-161

Rolling-Origin Evaluation Count

Problem

Count the number of rolling-origin forecasts: n - initial - horizon + 1 when positive, otherwise 0.

Each origin trains on a fixed window and forecasts horizon steps ahead.

Hint

Origins run from the initial window up to the point that still allows a full horizon.

case 1

in: [100,60,1]

expected: 40

case 2

in: [100,90,5]

expected: 6

case 3

in: [10,10,1]

expected: 0

case 4

in: [5,3,1]

expected: 2

Starter codePython

def rolling_origin_count(n, initial, horizon):
    # Your code here
    pass

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