Hi I am running a code in python importing rpy2 robjects and trying to make a fit to some data (with errors that I consider as weights). I am using a non linear regression and the "formula" i am trying to fit is: A/((x/t1)^b+(x/t1)^c)
Unfortunately every time I try to run the code I got the usual singular gradient error:
rpy2.rinterface.RRuntimeError: Error in function (formula, data = parent.frame(), start, control = nls.control(), : singular gradient
I think is due to my initial value for the parameters (A,t1,b and c). Is there any way I can have a better guess for the initial values instead just try-outs? Thanks, nino
