In: Statistics and Probability
Part 2. The table below provides the actual demand and a forecast for the last 12 months of Atwater's Problem Creation Company. Calculate the Cumulative Forecast Error and the Tracking Signal values for the last 12 months. Is the forecast Biased? Does it need to be replaced? | ||||||||||
Actual | New | Forecast | Abs. | Cumulative Abs | ||||||
Month | Demand | Forecast | Error | CFE | Error | Error | MAD | T.S. | ||
1 | January | 2000 | 1980 | |||||||
2 | February | 2400 | 2340 | |||||||
3 | March | 2300 | 2350 | |||||||
4 | April | 1800 | 1780 | |||||||
5 | May | 3300 | 3240 | |||||||
6 | June | 4425 | 4560 | |||||||
7 | July | 1900 | 1890 | |||||||
8 | August | 2000 | 2100 | |||||||
9 | September | 2200 | 1900 | |||||||
10 | October | 2200 | 1730 | |||||||
11 | November | 3325 | 3280 | |||||||
12 | December | 1700 | 1870 |
Let
At = Actual Demand
Ft = New Forecast
then
1. Forecast Error = At - Ft
2. The CFE (cumulative Forecast Error) values are obtained by adding the current Forecast error and the previous month's Forecast error values.
3. Absolute Error = |At - Ft|
4. The cumulative absolute error values are obtained by adding the current month's absolute error and the previous month's Forecast error values.
5. The MAD (Mean Absolute Deviation) value is obtained by taking the average of Absolute error values.
6. The Tracking Signal value is obtained by dividing the cumulative error by the MAD value.
From the table,
. | . | Actual | New | Forecast | Abs. | Cumulative Abs | . | . | |
Month | Demand | Forecast | Error | CFE | Error | Error | MAD | T.S. | |
1 | January | 2000 | 1980 | 20 | 20 | 20 | 20 | . | . |
2 | February | 2400 | 2340 | 60 | 80 | 60 | 80 | . | . |
3 | March | 2300 | 2350 | -50 | 30 | 50 | 130 | . | . |
4 | April | 1800 | 1780 | 20 | 50 | 20 | 150 | . | . |
5 | May | 3300 | 3240 | 60 | 110 | 60 | 210 | . | . |
6 | June | 4425 | 4560 | -135 | -25 | 135 | 345 | . | . |
7 | July | 1900 | 1890 | 10 | -15 | 10 | 355 | . | . |
8 | August | 2000 | 2100 | -100 | -115 | 100 | 455 | . | . |
9 | September | 2200 | 1900 | 300 | 185 | 300 | 755 | . | . |
10 | October | 2200 | 1730 | 470 | 655 | 470 | 1225 | . | . |
11 | November | 3325 | 3280 | 45 | 700 | 45 | 1270 | . | . |
12 | December | 1700 | 1870 | -170 | 530 | 170 | 1440 | . | . |
Sum | . | . | . | . | . | 1440 | . | . | . |
The tracking signal is positive which means the demand is greater than forecast.
Since the TS > 3.75, there is a forecast bias hence the forecast method needs to be replaced.