Re: question about the function 'fmins'

[Michael Creel]

For fitting by ML, I have some code in octave-forge that should do the trick. 
Just examine the file mle_example.m, in the main/econometrics subdir to see how 
to use it. Here's some sample output:

octave:1> mle_example


analytic score, unscaled data


******************************************************
Poisson MLE trial

MLE Estimation Results
BFGS convergence: Normal convergence

Average Log-L: -1.632551
Observations: 1000

         estimate     st. err      t-stat     p-value
beta1       1.013       0.011      93.503       0.000
beta2       1.994       0.015     133.783       0.000
beta3       2.996       0.002    1342.586       0.000

Information Criteria
CAIC : 3288.8256      Avg. CAIC:   3.2888
 BIC : 3285.8256       Avg. BIC:   3.2858
 AIC : 3271.1023       Avg. AIC:   3.2711
******************************************************

Cheers, Michael


Tracy Feldman wrote:
> To whom it may concern:
>  
> I recently downloaded Octave (I believe it is the most recent version) 
> onto a windows machine (2000? I am not sure, sorry), and am starting to 
> convert Matlab files (that had worked correctly) into Octave files.  I 
> am running into problems with the search algorithm 'fmins' (I had 
> previously used 'fminsearch' in Matlab, and I believe this is the 
> appropriate alternative in Octave?). 
>  
> Specifically, I wish to minimize a likelihood function, and I wish fmins 
> to return *both* the Maximum likelihood estimate (it is negative, so it 
> is actually a minimum) and a vector with the best fit parameters to the 
> model I am fitting (in this case, a logistic equation, binomially 
> distributed).
>  
> I tried doing it this way:
>  
> [BestFitParametersVector, MaximumLikelihoodEstimate] = 
> fmins(@LikelihoodFunction, InitialParametersVector, [], [], 
> TrialsVector, SuccessesVector, Density, LowerBoundsVector, 
> UpperBoundsVector);
>  
> But the program did not accept two elements in left side of the 
> equation, specifying what the function returns.  Is there a way to get 
> the function to return both of the things I want it to return?   I am 
> not sure how best to define these "returned" variables in the Likelihood 
> function itself (if this is necessary), or to specify that the function 
> return these variables?  
>  
> I would appreciate any ideas you might have on this.  Please let me know 
> if you need more information from me (or if more code might help).  If 
> possible, please respond directly to address@hidden 
> <mailto:address@hidden>, as I am not subscribed to the help 
> list.  Thank you in advance.
>  
> Sincerely,
>  
> Tracy S. Feldman
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://www.cae.wisc.edu/mailman/listinfo/help-octave