Cross-Learning Sample Count

Problem

Count training windows when multiple series are pooled for cross-learning: n_series * ((length - window) // stride + 1).

Return 0 when window or stride is not positive, or when length is shorter than the window. Pooling lets sporadic series borrow strength from each other.

Hint

Global models see the union of all series' windows during training.

Test cases

case 1

in: [10,100,24,1]

expected: 770

case 2

in: [3,50,10,5]

expected: 27

case 3

in: [5,10,10,1]

expected: 5

case 4

in: [2,10,11,1]

expected: 0

Starter codePython

def cross_learning_sample_count(n_series, length, window, stride):
    # Your code here
    pass

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