Estimation of linear and log-linear trend in the data and forecasting gasoline sales in US.
The following table shows gasoline sales in the United States (in millions of barrels) from the first quarter of 1986 to the last quarter of 1989.
GASOLINE SALES IN THE UNITED STATES: 1986:1 TO 1989:4
(IN MILLIONS OF BARRELS)
|
|
|
|
|
|
|
|
1986.1
|
600.4
|
1987.1
|
610.8
|
1988.1
|
640.8
|
1989.1
|
639.3
|
1986.2
|
652.8
|
1987.2
|
679.5
|
1988.2
|
686.5
|
1989.2
|
681.0
|
1986.3
|
671.1
|
1987.3
|
681.2
|
1988.3
|
691.1
|
1989.3
|
685.5
|
1986.4
|
654.8
|
1987.4
|
667.6
|
1988.4
|
676.5
|
1989.4
|
677.7
|
a. Approximate the linear trend in the data, also use it to forecast gasoline sales in the United States in each quarter of 1990.
b. Approximate the log-linear trend in the data, also use it to forecast gasoline sales in the United States in each quarter of 1990.
c.Which form of the trend fits the historical data better? Why would we expect both forecasts to be rather poor?