SES Alpha Fit Grid

Problem

Select the smoothing factor from a candidate grid that minimizes the one-step-ahead sum of squared errors. For each alpha initialize the level at x[0], then for t >= 1 accumulate (x[t] - s[t-1])^2 and update the level.

Return [best_alpha, best_sse]. If several alphas tie, keep the earliest candidate in the grid.

Hint

The forecast for x[t] is the level computed before seeing x[t].

Test cases

case 1

in: [[1,2,3,4],[0.1,0.5,0.9]]

expected: [0.9,3.4421]

case 2

in: [[5,5,5],[0.1,0.5,0.9]]

expected: [0.1,0]

case 3

in: [[1,3,5,7],[0.2,0.8]]

expected: [0.8,15.9104]

case 4

in: [[10],[0.3]]

expected: [0.3,0]

Starter codePython

def ses_alpha_fit(series, alphas):
    # Your code here
    pass

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