Page 1 of 1
maximum likelihood estimation
Posted: Wed Jul 07, 2010 1:05 am
by newbee
Hello, I need to estimate the parameters of a non-linear function simultaneously. I have the logistic function describing the cumulative probability of the observations but I do not know hot to use it in EViews (I have version 5). Can somebody help me? Thank you in advance.
Re: maximum likelihood estimation
Posted: Wed Jul 07, 2010 7:20 am
by EViews Gareth
You probably need to use the LogL object. The best thing to do is look it up in the User Guide or Help system, and give it a try.
Re: maximum likelihood estimation
Posted: Fri Jul 09, 2010 7:39 am
by newbee
I tried LOGL but when I ran the estimation I got the message "WARNING: Singular covariance - coefficients are not unique". I suppose the problem is in the starting values of the coefficients.
How can I bring EViews to run through different combinations of starting values for the parameters, make estimations and report the results?
Is there an easier solution?
Re: maximum likelihood estimation
Posted: Fri Jul 09, 2010 8:16 am
by EViews Gareth
You could write a program that loops through different values of starting values before estimation. Keeping track of which estimations "worked" and which didn't would be tricky though.
Re: maximum likelihood estimation
Posted: Fri Jul 09, 2010 12:25 pm
by newbee
Right, thank you for your help.