Simple Moving Average
Simple smoothing methods are weighted averages of past observations. Assume that the future will equal the averages of the past.
When seasonal patterns exist without trends, simple smoothing methods work well with deseasonalized data.
Forecast for a Simple Moving Average (SMA) is:
Ft+n = (Dt + Dt+1 + ... + Dn) / n
where:
t = time t
Dt = is the Demand Data for time t
n = the number of periods
The first forecast for n=4 and t=1 is:
Ft+n = (Dt + Dt+1 + ... + Dn) / n
Ft+n = F1+4 = F5 = (D1 + D2 + ... + D4) / 4
Demand 1: date 2000-01-01 value is 779
Demand 2: date 2000-02-01 value is 920
Demand 3: date 2000-03-01 value is 848
Demand 4: date 2000-04-01 value is 839
F5 = (779 + 920 + 848 + 839) / 4 = 846.5
The date for t=5 is 2000-05-01
We will make another forecast with n=8