I have a dataset that I'm trying to model with an equation. The points when connected produce a curve much like an upside down hockey stick:
I've been experimenting with an inverse square root equation y = b - a/sqrt(x). Below model uses a=100 and b=10. However I can't get the curve to turn sharply enough.
Am I on the right track? How can improve the fit?
Part of what was throwing me off was the upside-down-ness of the curve. This is also what was preventing Excel from coming up with a regression equation.
After realizing this, I made a new plot with the Y values subtracted from 10 which flipped it back over. Excel was then able to fit it to a power regression curve with equation y = a * x^b, where b was negative.