Croston's Method (forecast)

Question

I ran Croston's Method on this data(Sales) with h=3

    CustomerKey Period   Sales
    1      7869907      1    0.00
    2      7869907      2  418.62
    3      7869907      3   41.95
    4      7869907      4  306.40
    5      7869907      5    0.00
    6      7869907      6  151.46
    7      7869907      7  159.49
    8      7869907      8   49.72
    9      7869907      9   72.42
    10     7869907     10  240.48
    11     7869907     11  296.24
    12     7869907     12    0.00
    13     7869907     13  123.27
    14     7869907     14  294.76
    15     7869907     15    0.00
    16     7869907     16  180.80
    17     7869907     17   54.88
    18     7869907     18 2038.80
    19     7869907     19  187.22
    20     7869907     20  436.22

Applying fcast <- croston(y,h=3)

print(fcast)

    Point Forecast
    21       294.7139
    22       294.7139
    23       294.7139 

My question is that why there is no change in forecast for period 22 and 23 ?

Answer 1

Croston method and its variants create a linear prediction. It only changes when demand occurs.