In: Operations Management
Problem 6-07 (Algorithmic)
Refer to the gasoline sales time series data in the given table.
| Week | Sales (1000s of gallons) | 
| 1 | 17 | 
| 2 | 22 | 
| 3 | 19 | 
| 4 | 22 | 
| 5 | 19 | 
| 6 | 16 | 
| 7 | 22 | 
| 8 | 17 | 
| 9 | 23 | 
| 10 | 21 | 
| 11 | 14 | 
| 12 | 22 | 
Week  | 
Sales  | 
4 Period Moving Average  | 
5 Period Moving Average  | 
| 1 | 17 | ||
| 2 | 22 | ||
| 3 | 19 | ||
| 4 | 22 | ||
| 5 | 19 | ||
| 6 | 16 | ||
| 7 | 22 | ||
| 8 | 17 | ||
| 9 | 23 | ||
| 10 | 21 | ||
| 11 | 14 | ||
| 12 | 22 | 
Please refer below table .
We place all the data provided in excel table
We calculate four week and five week moving average as per following formula :
Four week moving average :
Ft = ( At-1 + At-2 + At-3 + A t-4) / 4
Five week moving average :
Ft = ( At-1 + At-2 + At-3 + A t-4+ At-5 ) / 5
Ft = forecast for week t, At , At-1 etc = Actual data for period t, t-1 etc
Squared error = Calculate square of absolute difference forecast data and actual data
| 
 Week  | 
 Actual data  | 
 Four week moving average  | 
 Squared error  | 
 Five week moving average  | 
 Squared error  | 
| 
 1  | 
 17  | 
||||
| 
 2  | 
 22  | 
||||
| 
 3  | 
 19  | 
||||
| 
 4  | 
 22  | 
||||
| 
 5  | 
 19  | 
 20  | 
 1  | 
||
| 
 6  | 
 16  | 
 20.5  | 
 20.25  | 
 19.8  | 
 14.44  | 
| 
 7  | 
 22  | 
 19  | 
 9.00  | 
 19.6  | 
 5.76  | 
| 
 8  | 
 17  | 
 19.75  | 
 7.56  | 
 19.6  | 
 6.76  | 
| 
 9  | 
 23  | 
 18.5  | 
 20.25  | 
 19.2  | 
 14.44  | 
| 
 10  | 
 21  | 
 19.5  | 
 2.25  | 
 19.4  | 
 2.56  | 
| 
 11  | 
 14  | 
 20.75  | 
 45.56  | 
 19.8  | 
 33.64  | 
| 
 12  | 
 22  | 
 18.75  | 
 10.56  | 
 19.4  | 
 6.76  | 
| 
 Total =  | 
 116.44  | 
 84.36  | 
|||
| 
 Mean Square error (MSE)  | 
 14.55  | 
 12.05  | 
Basis above,
Mean squared error (MSE) :
For 4 week moving average case = Sum of squared error/ 8 i.e. number of data = 14.55
For 5 week moving average case = Sum of squared error /7 i.e. number of data = 12.05
MSE for 3 week moving average = 12.1138
A lower MSE is always desirable since it indicates least difference between actual and forecast data.
Hence 5 week moving average case is the best and is to be used for moving average calculation