inaccurate coefficients with Polynomial fit function

[mbeuret]

Hello,

I don’t understand how using the General Polynomial fit function. I found inaccurate coefficients  (I’m beginner with Labview)

I need to calculate the coefficients (A, B, C, D) for this equation:  Y = A+BX+CX^2+DX^3.

With Excel I found 7 points to test the coefficient resolution with the Regression Solver.vi

The Regression Solver Results are different if I give 4, 5, 6 or 7 knowing points !!!

4 pts =>  y = -301.785E+0 + 4.261E+0x - 16.342E-6x^2 + 16.001E-12x^3  
5 pts =>  y = +491.811E+3 - 380.684E-3x - 1.888E-6x^2 + 1.137E-12x^3 
6 pts =>  y = +491.694E+3 - 379.525E-3x - 1.892E-6x^2 + 1.140E-12x^3

7 pts =>  y = -106.524E+3 + 2.334E+0x - 1.945E-6x^2 + 3.032E-12x^3 

Why with 4 and 7 points the coefficients are wrong?

For helping you how I generate those results, you will found an Excel File and the VI in attachment.

Thanks a lot for your helps

Mark
 

[mbeuret · Accepted solution]

I have the solution.

I use an another algorithm !!   Only one algorithm   give wrong results in a particular way and it is the default: SVD 

Mark