If you haven't been told what type of function (e.g. linear, quadratic, power, exponetial, etc) to model the data with, you can use plot points manually or do it on a graphics calculator. I have a CASIO fx-9750, and I suspect you have a Ti (simply because I always see them mentioned here) , but for the casio I go to the Stat menu and enter the years - if x=0 at 1970, you have x=0,5,15...30 - into the list 1 and the percentages into list 2.
It looks linear and the correlation coefficient, r, for a straight-line model is 0.9925 and is the most reasonable.
If you do it manually, you would need to estimate the line of best fit, and find the gradient of that line (don't just do (last y value - first)/(last x - first x) ).
The graphics calculator tells me
y = 0.8064x + 10.59 **
or f(x) = 0.8064x + 10.59
One way to find the inverse is to swap x and y in ** which gives
x = 0.8064y + 10.59
And putting it back into the standard form y = mx + c
y = (x-10.59)/0.8064
y = 1.24x - 13.1
or f<sup>-1</sup>(x) = 1.24x - 13.1
If you're using a graphics calculator, you could now use its graphing function to graph these two lines. Because they are inverses, they should be mirrow images across the line y=x.
I hope that helps you
.
