Further Questions

[kimk2kr]

I just have some questions about time series, finding trend line and deseaonlising the time series

year        CO₂ Emission
1972       0.33
1973       0.33
1974       0.34
1975       0.36
1976       0.36
1977       0.42
1978       0.41
1979       0.42
1980       0.44
1981       0.46
1982       0.48
1983       0.51

so questions i have to ask is.
1) why is it not necessary to deseasonlise the data before you put in a trend line?
2) How to find a trend line by hand without using any sort of technology using only 2 coordinate?
and lastly 3) if we were to find the CO₂ emission in the year 1985 ad 1988 why would extrapolating data be less accurate?

thanks!!

[amilss]

1) well you deseasonalise the data to get rid of any seasonal variations. Because it's a yearly value i would assume there is no seasonal variation

2)- pick two points off a line of best fit (x1,y1) and (x2,y2)
- find the gradient of the trend line m=(y2-y1) divided by (x2-x1)
-then sub that into y-y1=m*(x-x1) to get the y-intercept(aka c)
- then sub those values into the equation form y=mx + c

3) i assume you mean 1985 and 1988 are predicted using the trend line. Well it's just a line of best fit so the trend equation is not the most accurate measure of the relationship between variables to begin with. If the data is extrapolated it probably means the trend line is not the best fit that it could be. I dont really know the answer to the last Q sorry

[kimk2kr]

for finding the trend line do you just pick ANY two points?

thanks for the help

[amilss]

Yes any two points on the line of the best fit that you draw in

[kimk2kr]

ahhhh right!! i get it! thanks so much