Fit a trend line to the data in Problem 7 by the method of least squares. Also, obtain the trend value for the year 1990.
Solution
In the given problem, n = 11 (odd), middle t- values is 1981, h = 1
u = `"t - middle value"/"h" = ("t" - 1981)/(1)` = t – 1981
We obtain the following table.
Year t |
Production yt |
u = t–1981 | u2 | uyt | Trend Value |
1976 | 0 | –5 | 25 | 0 | 1.6819 |
1977 | 4 | –4 | 16 | –16 | 2.4728 |
1978 | 4 | –3 | 9 | –12 | 3.2637 |
1979 | 2 | –2 | 4 | –4 | 4.0546 |
1980 | 6 | –1 | 1 | –6 | 4.8455 |
1981 | 8 | 0 | 0 | 0 | 5.6364 |
1982 | 5 | 1 | 1 | 5 | 6.4273 |
1983 | 9 | 2 | 4 | 18 | 7.2182 |
1984 | 4 | 3 | 9 | 12 | 8.0091 |
1985 | 10 | 4 | 16 | 40 | 8.8 |
1986 | 10 | 5 | 25 | 50 | 9.5909 |
Total | 62 | 0 | 110 | 87 |
From the table, n = 11, `sumy_"t" = 62, sumu = 0, sumu^2 = 110, sumuy_"t" = 87`
The two normal equations are : `sumy_"t" = "na"' + "b"' sumu "and" sumuy_"t" = "a"' sumu + "b"'sumu^2`
∴ 62 = 11a' + b'(0) ...(i) and
87 = a'(0) + b'(110) ...(ii)
From (i), a' = `(62)/(11)` = 5.6364
From (ii), b' = `(87)/(110)` = 0.7909
∴ The equation of the trend line is yt = a' + b'u
i.e., yt = 5.6364 + 0.7909 u, where u = t – 1981
∴ Now, For t = 1990, u = 1990 – 1981= 9
∴ yt = 5.6364 + 0.7909 x 9 = 12.7545.