Maximum Likelihood Estimation

38 views
Skip to first unread message

ahmad.jal...@gmail.com

unread,
Jan 8, 2013, 6:18:05 AM1/8/13
to accor...@googlegroups.com
Hi Cesar,

I have a given model and I have to estimate the MLE of this model in five dimensional parameter space.

I have implemented my own Brute Force Maxima determination method to estimate the parameter values. Good thing is that its working but sometimes its very slow because my input data is very big (millions or records). I was wondering if there is some implementation of MLE in Accord.NET framework.

Many thanks,

Ahmad Qarshi

mu li

unread,
Jan 8, 2013, 6:36:39 AM1/8/13
to accor...@googlegroups.com
Hi,

What kind of distribution you want to MLE?
In my option if the distribution is Gaussian or one of the exponential families the MLE is very easy with using sufficient statistics. 
If you are using more complex distribution, the computation always slow, you can try Gibbs sampling or some approximation method.

Sorry for my interruption of your question to Cesar.
Reply all
Reply to author
Forward
0 new messages