Question

15. Two methods are used to predict how many customers will call in for help in the next four days. The first method predicts the numbers of callers to be 23, 5, 14, and 20 for the four respective days. The second method predicts 20, 13, 14, and 20 for the four respective days. The actual numbers of callers turn out to be 23, 10, 15, and 19. Which method has the bigger forecast bias

Answer 1

Method 1 has bigger forecast bias.

Explanation:

We are given the following information in the question:

Actuals:

23, 10, 15, 19

Method 1 prediction:

23, 5, 14, 20

Residuals = Predicted - Actuals

Sum of square of errors:

`=(23-23)^2+(5-10)^2+(14-15)^2+(20-19)^2\\= 27`

Mean squared error:

`\frac{\text{Sum of squares}}{n} = (27)/(4) = 6.75`

Method 2 prediction:

20, 13, 14, 20

Residuals = Predicted - Actuals

Sum of square of errors:

`=(20-23)^2+(13-10)^2+(14-15)^2+(20-19)^2\\= 20`

Mean squared error:

`\frac{\text{Sum of squares}}{n} = (20)/(4) = 5`

Since, mean square error of method 2 is less as compared to method 1, method 1 has bigger forecast bias.