Seasonal Naive MAE

Problem

Compute the in-sample mean absolute error of the seasonal naive method with period m. The error at time t is |x[t] - x[t-m]|, averaged over every t that has a full season of history.

Return 0.0 when m is not positive or when the series is not longer than one season.

Hint

There are n - m seasonal naive errors in a series of length n.

Test cases

case 1

in: [[1,5,2,6,3,7],2]

expected: 1

case 2

in: [[2,4,2,4],2]

expected: 0

case 3

in: [[1,2,3],5]

expected: 0

case 4

in: [[5,7,9],1]

expected: 2

Starter codePython

def seasonal_naive_mae(series, m):
    # Your code here
    pass

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