The simple mobile forecasting method is used when you want to give more importance to newer datasets to get the forecast. Each point of a moving average of a time series is the arithmetic mean of a number of consecutive points in the series, where the number of points is chosen in such a way that seasonal and/or irregular effects are eliminated.
When to use a moving average forecast?
The moving average forecast is optimal for random or level demand patterns where the impact of historical irregular elements is intended to be eliminated through a focus on recent demand periods.
Moving Average Model
Formula
Average sales in units in the period “t“
Data summaries
Actual sales in units of periods prior to “t“
Number of data
Example of applying a Mobile Average forecast
A company presents in the next tabulated sales report for the year 2017.
| Month | REAL SALES (2017) |
| January | 80 |
| February | 90 |
| March | 85 |
| April | 70 |
| May | 80 |
| June | 105 |
| July | 100 |
| August | 105 |
| September | 100 |
| October | 105 |
| November | 100 |
| December | 150 |
Taking into account the above data, a forecast should be calculated using the Moving Average technique using:
- A period of 3 months (as of April 2017)
- A 6-month period (as of July 2017)
The goal is to identify which of the two forecast periods is most accurate when compared to the actual sales of the report.
Solution
As it is a forecast with a mobile period of 3 months, it must be made from april, that is, for its calculation it will take into account three periods, i.e. January, February and March.
Then to make the forecast for the month of May, the last three periods preceding the month of May, i.e. February, March and April, must be taken into account.
In this way the remaining forecasts are made obtaining the following result:
| Month | REAL SALES (2017) | PROGNOSIS 3 MONTHS |
| January | 80 | |
| February | 90 | |
| March | 85 | |
| April | 70 | 85 |
| May | 80 | 82 |
| June | 105 | 78 |
| July | 100 | 85 |
| August | 105 | 95 |
| September | 100 | 103 |
| October | 105 | 102 |
| November | 100 | 103 |
| December | 150 | 102 |
The remaining forecast being a forecast with a moving period of 6 months, this must be made from the month of July, that is to say for its calculation will take into account six periods, i.e. January, February, March, April, May and June.
In this way the remaining forecasts are made obtaining the following result:
| Month | REAL SALES (2017) | PROGNOSIS 3 MONTHS | FORECAST 6 MONTHS |
| January | 80 | ||
| February | 90 | ||
| March | 85 | ||
| April | 70 | 85 | |
| May | 80 | 82 | |
| June | 105 | 78 | |
| July | 100 | 85 | 85 |
| August | 105 | 95 | 88 |
| September | 100 | 103 | 91 |
| October | 105 | 102 | 93 |
| November | 100 | 103 | 99 |
| December | 150 | 102 | 103 |
Although there are several accuracy indicators of a forecast, in this case the result is more than obvious, since we can see how the forecast with a 3-month moving period manages to approximate to a greater extent the actual sales of 2017 in relation to the forecasts obtained by forecasting with a 6-month moving period.
