Fitting some random data into beta distribution

53 views
Skip to first unread message

Shekhar

unread,
May 3, 2011, 7:39:06 AM5/3/11
to Bangalore R Users - BRU
Hi,
I have some random data, and i want to find the parameters of Beta
distribution ( a and b) for which this random data approximately fits.
I have tried by plotting graphs but that is not very intutive and
requires lot of tuning. Can some body provide me some pointers??

Regards,
Som Shekhar

Varun Varma

unread,
Sep 1, 2011, 10:36:37 AM9/1/11
to bru...@googlegroups.com
I do not know if you have found an answer to this, but a crude way of doing this would be to iteratively generate random data using beta distributions where a and b vary. Then test each generated distribution against your random data using a Kolmogorov-Smirnov test.

The Kolmogorov-Smirnov test is called using ks.test([your data], [generated data OR a cumulative distribution function])

This should at least give you a range of values for a and b within which your data falls.


Hope this helps,
Varun
Reply all
Reply to author
Forward
0 new messages