The Fisher Information (FI) has been widely applied to ecological, economic and social systems to analyze dynamic order in multivariate data, including regimes and abrupt transitions (aka regime shifts). While software in literature to compute FI include MATLAB and Phyton (https://github.com/csunlab/fisher-information), I'm wondering if there are available R packages/funtions for calculation. Thanks!
Xavier
I've not used the abovementioned package.
However, an aleternative option to calculate the FI could be the nlm function (non linear minimization):
https://www.stat.umn.edu/geyer/5931/mle/mle.pdf
Hi Dear Xavier Benito-Granell
To compute the elements of expected Fisher information matrix, I suggest to use Variance-Covariance matrix as in vcov( ) function by 'maxLik' package in R, the inverting vcov( )^-1, to return the Fisher information matrix.
This package is easily in computations, see attached file.
Regards
Dear Friends
Dear dr Ahmed Elshahhat
I support your words, but I have a small objection
To compute Variance-Covariance matrix in R program by ('maxLik' or 'bbmle') package in R use "vcov(fit)" .
The inversion of Variance-Covariance matrix is Fisher information matrix.
In R program don't use "vcov(fit)^-1" to compute Fisher information matrix, but use "solve(fit)".
Best regards
Hi Dear Dr. Ehab M. Almetwally
Already the command vcov^-1 it is known as inverting V-Cov matrix by the command solve(vcov) in R.
Regards
The Observed Fisher Information is the Hessian matrix for likelihood function in the computational part of any optimizing tool. The nlm or optim functions in R provide hessian matrix if we optimize log-likelihood function and that will be observed Fisher Information matrix.
- Badges
- Science topic