Question

The following table displays the actual demand and forecasts using two alternative methods for a six-month period.

		FORECAST
Month	Actual	Method 1	Method 2
1	161	152	154
2	163	164	155
3	151	163	151
4	162	161	152
5	156	163	152
6	164	163	154

Compute MSE for both methods. Using MSE as a criterion, which method has the better performance record and why?

Answer 1

Method 1:

Month	Actual	Forecast	Error (Actual - Forecast)	Error^2
1	161	152	9	81
2	163	164	-1	1
3	151	163	-12	144
4	162	161	1	1
5	156	163	-7	49
6	164	163	1	1

MSE (Mean Squared Error) = Mean of (Error^2) = ( 81 + 1 + 144 + 1 + 49 + 1) / 6 = 277 / 6 = 46.167

Method 2:

Month	Actual	Forecast	Error (Actual - Forecast)	Error^2
1	161	154	7	49
2	163	155	8	64
3	151	151	0	0
4	162	152	10	100
5	156	152	4	16
6	164	154	10	100

MSE = Mean of (Error^2) = ( 49 + 64 + 0 + 100 + 16 + 100) / 6 = 329 / 6 = 54.833

MSE for Method 1 is lower.

Therefore, Method 1 has a better performance record using MSE as a criterion.