Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Parameter estimation

  1. Jun 25, 2007 #1
    I have a probability distribution of the form [tex]\sum_{i=0}^n f(n,x,y)[/tex]. There is no closed form expression for it. I need to know if there is any method that I can use to estimate the parameters {n, x, y} given some data from the above distribution.
    I've tried a maximum likelihood approach, but I'm having trouble getting the derivative with respect to n. Is it possible to get this derivative, and use a maximum likelihood approch to estimate n
     
  2. jcsd
  3. Jun 25, 2007 #2

    EnumaElish

    User Avatar
    Science Advisor
    Homework Helper

    Is your f indexed by i? If not, then you have (n+1)f(n,x,y), which is differentiable w/r/t/ n, as long as f is.

    If f is indexed by i, then you might think of the sum as an integral and may be able to apply Leibniz's Rule (see under "Alternate form": http://en.wikipedia.org/wiki/Leibniz's_rule).
     
    Last edited: Jun 25, 2007
  4. Jun 28, 2007 #3
    Thanks for the reply. I had a look at that Leibniz's Rule link, but I'm not fully sure how to go about using it??

    Anyway, I was thinking of a slightly more simple idea. I basically need an estimate of the 3 parameters {n, x, y}, preferiably using MLE. Since it's difficult to get the derivative w.r.t n, I was thinking of trying various values of n (say n=1,..,50), and for each value of n estimate MLE of x,y.

    So basically, I now end up with 50 different estimates for {n, x, y}. So my question is, is there any mathematical way to tell which one of these 50 estimates is the best one? ie. Is there some sort of likelihood test I could use?
     
  5. Jun 28, 2007 #4

    EnumaElish

    User Avatar
    Science Advisor
    Homework Helper

    I'd just look at the (log) likelihood numbers and select the largest.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?