Can't reproduce original data from curve fit function

I am using matlab cftool to fit a X vs Y set of data. Both X and Y are 1D arrays. The curve generated by the cftool visually fits the data well, and if I use the 'analysis' section of the cftool, I get Y_fit very close to Y. But if I get the fitting coefficients are generate Y_fit from them, the values are way off. Am I making any basic mistake? I am using polynomial curve fit scheme.

[Sean de Wolski]

How are you getting the coefficients?

Save the fit to workspace "Fit Menu" -> "Save to Workspace"

Then

fittedmodel.C

where C is your coefficient

[Ram]

The coefficients are displayed with 95% confidence bounds in the Fit Result window of the cftool application. I tried the way you suggested above and these are the exact coefficients I was using. I've attached my data sheet (res vs temp), and here are the results from fit:

Linear model Poly5:

     f(x) = p1*x^5 + p2*x^4 + p3*x^3 + p4*x^2 + p5*x + p6
Coefficients (with 95% confidence bounds):

       p1 =      -22.29  (-27.5, -17.08)
       p2 =       356.9  (280.5, 433.2)
       p3 =       -2321  (-2766, -1875)
       p4 =        7696  (6405, 8988)
       p5 = -1.308e+004  (-1.495e+004, -1.122e+004)
       p6 =        9165  (8094, 1.024e+004)

Goodness of fit:

SSE: 0.4659
R-square: 1
Adjusted R-square: 1
RMSE: 0.08739

When I use these coefficients to calculate the values back, they are very much off from those generated by the 'analysis' section in cftool application (.fig attached)