I came across many statistical packages such as MATLAB which has Distribution fitting toobox which employs log likelihood method for estimation of parameters, EasyFit tool which fits many distribution along with that it provides goodness of fit measures too, also CumFreq software which enables probability fitting of a limited number of distributions.
As all these packages involve different methods of parameter estimation and have dissimilar list of inbuilt probability distribution, a comparison of these will not be worthy.
Kindly suggest the most suitable and acceptable solution!!
Regards!
I am most familiar with R, and I think it can do what you want. However, there might not be enough tools in base R. You might need to add additional packages.
Base R comes with support for generalized linear models with the glm() function, which is quite powerful. There are only a few distribution families to choose from "out of the box," including Poisson, Gaussian, inverse Gaussian, Gamma, and a couple of others. However, you can expand R's list of available distribitions by adding certain other packages (I don't have a link right now, but Googling will get you there).
You can also attempt to find a good-fitting distribution for your data using a variety of methods. Me, I like to just run the GLM model with the distribution in question then look at the residuals to see whether the distribution was a good choice (anyone can feel free to tell me if this is a bad idea). However, there are several functions in the package called fitdistrplus that can help, as well.
All of this assumes you have basic (and maybe a bit more) proficiency with R.
http://www.statmethods.net/advstats/glm.html
https://cran.r-project.org/web/packages/fitdistrplus/fitdistrplus.pdf
Priyank... R is the best & most user friendly programming platform where you can simply do maneuver to anything related to statistics
The best applications for statistical modeling are SPSS, STATISTICA, MATLAB and R. Nevertheless, if you want to have a really good control of your work the best idea is to make your own modeling of the distributions. As you have said, each software implements a different method for the estimation of parameters and uses different notations.
If you have an idea of what distribution may relate better to your data, I could give you some recommendations of bibliography.
Good Luck!
You may use Easy fit software which has alarge number of frequency distribution and provides detailed report.
Could you please tell me the software or python libarary for probability distribution fitting like easyfit
Priyank J. Sharma , I found Distfit and Fitter packages for different distribution fitting using distinct metrics like residual sum square(RSS) for prior and sum square error (SS)
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning