Croston vs SES Comparison

Problem

Compare Croston and simple exponential smoothing on the same intermittent series with smoothing factor alpha. Return [croston_forecast, ses_forecast], where the SES level is initialized with the first observation and updated over all points, including zeros.

If the series has no nonzero demand the Croston forecast is 0.0.

Hint

Croston ignores zero periods when updating demand size; SES does not.

Test cases

case 1

in: [[0,0,4,0,0,6,0,2],0.5]

expected: [1.4,1.8125]

case 2

in: [[0,0,0],0.3]

expected: [0,0]

case 3

in: [[5],0.5]

expected: [5,5]

case 4

in: [[1,0,3,0],0.5]

expected: [1.3333333333333333,0.875]

Starter codePython

def croston_vs_ses(series, alpha):
    # Your code here
    pass

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