Alternative to curve_fit() from scipy.optimize

ddidussa at ddidussa at
Sun Apr 19 22:56:50 CEST 2015


I am trying to fit a curve of the form (ln(x+a))**b to a set of points.

However, curve_fit() from scipy.optimize fails to find a consistent optimal solution (as I increase the number of data points, the coefficients found vary greatly).

I suspect this is because of the algorithm that the function uses to solve the problem (Levenberg-Marquardt), as it finds a local minimum and returns those values.

Do you know of any package that contains a function that is more adequate to fit logarithmic functions? Or a curve-fitting function where I can specify which algorithm to solve the problem?

Thanks in advance,

More information about the Python-list mailing list