Using icdf() function with custom distribution

4 vues (au cours des 30 derniers jours)
Aaron J. Hendrickson
Aaron J. Hendrickson le 15 Mar 2017
According to MathWorks here :
"User-defined custom distributions, created using files and function handles, are supported by the Statistics and Machine Learning Toolbox™ functions pdf, cdf, icdf, and mle, and the Statistics and Machine Learning Toolbox Distribution Fitter app."
Details of problem:
I have a rather complicated probability distribution that I am working with. I have provided a picture of the pdf and cdf below. I need to find the quantile function by inverting the cdf. I have already done this by writing a function handle for the cdf and then inverting it using fzero. This works but its slow and is not an ideal method for numerically inverting a cdf. Therefore, I believe this is not the best approach to take. On the other hand, the icdf function probably uses more appropriate techniques for inverting cdf's and I would like to use that instead.
My question:
If I understand the above statement from mathworks, I should be able to use the icdf function given a function handle for my distribution. I have tried to make this work but with no success. How can one create/define a custom distribution such as mine and use it with the statistics toolbox functions above? Any examples would be helpful. I can certainly provide my code if requested as well. Also, I do not need the ability to fit data using the distribution fitting tool in case that matters.
Thanks.

Réponses (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by