Question. 1.
a. The sales of Diesel Generating sets for M/S Bright Generators Ltd. are given in the
following table for the last 12 years. The figures are in thousand numbers.
Use the least square regression analysis to forecast the sales of generating sets for the
years 2007 and 2008.
b. For the above data assume that the forecast for the year 1995 was 2100. If the
smoothing factor is 0.4 project the forecasts for the years 1996 to 2006 using exponential
smoothing method.
c. Using Simple Moving Average of three years, project the sales of Generating sets for
the years 1998 to 2006 for the above example.
Question 2.
The actual sales figures of air conditioners manufactured by M/S Cool Safe Ltd. are
given for the past 10 years in the following Table. The sales figures are in thousand
numbers. As a marketing Consultant for the company, you have to forecast the sales for
the year 1911 using two years and three years simple moving average method.
YR ACTUAL YR ACTUAL
SALES SALES
1901 230 1906 260
1902 220 1907 300
1903 200 1908 240
1904 240 1909 280
1905 230 1910 320
Which of these forecasts would be more reliable based on “Mean Squared Error” basis?
SOLUTION 1:
Yt = a + bT
∑TY–nTY
b=
∑T2–nT2
a= Y – bT
T Y TY T2
1 (1995) 2,000 2,000 1
2 2,200 4,400 4
3 2,100 6,300 9
4 2,300 9,200 16
5 2,500 12,500 25
6 3,000 18,000 36
7 3,200 22,400 49
8 3,500 28,000 64
9 4,000 36,000 81
10 3,500 35,000 100
11 4,500 49,500 121
12 (2006) 5,000 60,000 144
53506
= = 259.23
206.4
a = Y – bT
= 3,150 – 259.23 (6.08) = 1756.12
Thus linear regression will be
Y = 1756.12 + 259.23 T
Hence the forecast for the year 2007 (13th year) will be
Since the numbers are in thousands, the sales forecast for the year 2007 will be
5,126,000 Generating sets.
Similarly, the forecast for the year 2008 (14th year) will be
Since the numbers are in thousands, the sales forecast for the year 2008 will be
5,385,000 Generating sets.
The forecasts for periods 2 (1996) to 12 (2006) can be then calculated as follows:
1 2,000
2 2,200
3 2,100 F4 = (2000 + 2200 + 2100)/3 = 2100
4 2,300 2100 F5 =(2200 + 2100 + 2300)/3= 2200
5 2,500 2200 F6 = (2100 + 2300 + 2500)/3 = 2300
6 3,000 2300 F7 = (2300 + 2500 + 3000)/3= 2600
7 3,200 2600 F8 = (2500 + 3000 + 3200)/3 = 2900
8 3,500 2900 F9 = (3000 + 3200 + 3500)/3 = 3233
9 4,000 3233 F10 = (3200 + 3500 + 4000)/3 = 3567
10 3500 3567 F11 = (3500 + 4000 + 3500)/3 =3667
11 4500 3667 F12 =(4000 + 3500 + 4500)/3 = 4000
12 5,000 4000 F13 = (3500 + 4500 + 5000)/3 = 4333
13 - 4333
SOLUTION 2.
Answers:
1) Forecast of sales in the year 1911 based on 2 years moving average = 300,000
2) Forecast of sales in the year 1911 based on 3 years moving average = 280,000
3) Mean Squared Error being less for the two year moving average, forecast with two
year moving average is more reliable.
(Note: Marks could be deducted if the units are not mentioned in the answer )