How to reduce the rmse for a fit obtained using curve fitting toolbox

Hi I need to find a function which best fits x and y data points as given below

xdata=[0.500000000000000,0.700000000000000,1];
ydata=[7862.13480751439,5801.18185697929,4993.13280898239];

When I use Curve fitting toolbox, it returns a polynomial function with the goodness of fit as given below

Linear model Poly2:
     f(x) = p1*x^2 + p2*x + p3
Coefficients:
       p1 =   1.522e+04
       p2 =  -2.857e+04
       p3 =   1.834e+04

Goodness of fit:
  SSE: 1.414e-22
  R-square: 1
  Adjusted R-square: NaN
  RMSE: NaN

ydata points appears to coincide the curve in the plot and also the sse and Rsquare are desirable but the rmse returned by the goodness of fit is NaN. Can this rmse be negelected? How is Adjusted R-square value calculated?

Answer 1

[Ahmet Cecen]

Read up on the concepts of Overfitting, Underfitting, Variance and Regression. You are fitting a function of 3 variables to 3 data points. I would say a regression problem with 3 data points is fairly meaningless to begin with, but if you have to do it, fit a line instead.