In my research, I developed a model (for a stochastic scheduling problem) which works with stochastic variables. These stochastic variables assumed to follow phase type distribution in the model. One of the databases in the literature, however, contains beta distributed data, which theoretically cannot be solved using our model.
We think it might be possible to approximate these beta distributed data by a phase type distribution. I have a limitation on the number of moments (should be less than 5).
What is the easiest and/or best way of doing that?
Thanks,
.
an example is given in the link below (Section 2.2.1, with only two moments)
http://www.ise.ncsu.edu/people/faculty/docs/IETrans.pdf
solving these non-linear problems is not "easy" ...
.
.
According to Wikipedia, phase-type distributions can be used to approximate any positive-valued distribution.
- Badges
- Science topic