Question

The number of heart surgeries performed at Heartville General Hospital has increased steadily over the past several years. The hospital's administration is seeking the best method to forecast the demand for such surgeries in year 6. The data for the past five years are shown below. 

Year	1	2	3	4	5
Demand	44	47	54	57	61

The hospital's administration is considering the following forecasting methods. Begin error measurement in year 3, so all methods are compared for the same years 

i. Exponential smoothing, with a 0.6. Let the initial forecast for year 1 be 44, the same as the actual demand. 

ii, Exponential smoothing, with o: 0.9 Let the initial forecast for year 1 be 44, the same as the actual demand. 

iii. Trend projection with regression 

iv. Two-year moving average.

 v. Two-year weighted moving average, using weights 0.6 and 0.4, with the more recent data given more weight. 

If MAD is the performance criterion chosen by the administration, which forecasting method should it choose? Trend projection with regression

 If MSE is the performance criterion chosen by the administration, which forecasting method should it choose?

Answer 1

Answer: according to MAD forecast -Trend projection regression (MAD value is 0.600)

Answer: if MSE is criteria- trend projection regression (because MSE is 0.707, and none of the forecast's MSE will come less than '1')

Explanation:

Absolute deviation= |Forecast - Sales|

MAD= mean absolute deviation= sum of absolute deviation/no. of months

• the formula to be used in Exponential smoothing is

F_t+1= alpha*A_t + (1-alpha) F_t

A_t means Actual demand of t'th month, if you want to find out the Forecast through exponential smoothing= forecast of 3rd month = alpha*actual demand of 2nd month+(1-alpha) *forecast demand of 2nd month

remember forecast of 1st month is actual demand of 1st month

• Forecast of a month with two period moving average = (sales of most recent month+ sales of next most recent month)/2
• Weighted Forecast of a month = (0.6*sales of most recent month+ 0.4*sales of next most recent month)