Expanding Window Splits

Problem

Generate expanding-window time series splits. For each t from initial_train to n-1, the training set is range(t) and the test set is [t].

Return the list of [train_indices, test_indices]; if initial_train is negative or at least n, return [].

Hint

The training window grows by one sample as the test point moves forward.

case 1

in: [5,3]

expected: [[[0,1,2],[3]],[[0,1,2,3],[4]]]

case 2

in: [4,1]

expected: [[[0],[1]],[[0,1],[2]],[[0,1,2],[3]]]

case 3

in: [3,3]

expected: []

case 4

in: [3,4]

expected: []

Starter codePython

def expanding_window_splits(n, initial_train):
    # Returns list of [train_indices, test_indices]
    # Your code here
    pass

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