Hello,
I am using Excel Solver to resolve the best fit of a polynomial function to a known data set by altering the coefficients (weights) of the polynomial. The problem I have is most of the weights are resolved to zero and this producing a flat line (which does not fit the data well at all). What could be the cause of this?
For instance, in the polynomial function, y=(w*(0))+(w*(1)*(x))+(w*(2)*(x^2)), I have set the Objective cell to the sum of the squared error value and the weights, w*(0), w*(1), and w*(2), as the Changing Variable cells. The weights are set to values that approximately fit the known data set. When I run solver, I indicate I want a min value of the objective cell. The solution solver provides is w*(0)=non zero value and w*(1) and w*(2)= zero.
I have another sheet in excel in which I am performing the exact form of fitting but with a cubic polynomial, y=(w*(0))+(w*(1)*(x))+(w*(2)*(x^2))+(w*(3)*(x^3)).
Attached is the excel file.
I appreciate any feedback
Bookmarks