Question 1137344
Linear=23.58-0.01x with an R^2 of about 0,, which is essentially no correlation at all
Quadratic is -0.457x^2+3.647x+19.92 with an R^2 of 0.999, which is almost no deviation at all from a quadratic model (not as surprising given few data points and a rise and a fall).

-0.457x^2+3.647x+19.92=25.0
-0.457x^2+3.647x-5.08=0
x+(-1/0.914)(-3.647 +/- sqrt 4.01); sqrt is 2

x=(-1/.914)(-5.647)=6.18 or about 2 months into 1997 it was 25 million
x=(-1/.914)(-1.647)=1.80 or about 10 months into 1992 it was 25 million

in 1995, it was 26.73, letting x=5

The y-intercept is the number in 1990.
The x-intercept would be further out in time or further back in time and might correspond to when the program began or not.  It is beyond the capability of the model.

Practical domain is within 1990-1998
practical range is between 19.8 and 27.3, since we have no further data to support anything further.

Maximum is when x=-b/2a or 3.99
f(x)=27.19 at almost exactly 4.0, despite the actual data's showing 27.3 at 4.0


{{{graph(300,300,-3,10,-10,30,-0.457x^2+3.647x+19.92)}}}