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

Estimating parameters from multivariate normal

  1. May 17, 2008 #1
    I need to estimate parameters from data that follow a mutinormal distribution. The parameters that I need to estimate are contained in the expression for the mean of the marginal normal distributions. That is each marginal distribution has mean:

    [tex] \frac{p_1*c_i + p_1*y}{p_1+p_2}[/tex]

    where [tex]p_1[/tex], [tex]p_2[/tex] and [tex]y[/tex] are the paramters that I need to estimate and [tex]c_i[/tex] is just a known contant associated with the ith marginal. [tex]p_1[/tex] and [tex]p_2[/tex] are random parameters and [tex]y[/tex] is a fixed parameter.

    I've tried using a multinormal likelihood approach, which gives an estimate for the mean as the sample mean, but how do I get estimate of the actual parameters from this sample mean? Can I even use this approach?

    Also is there an added complexity to estimating parameters when there are random parameters, as is the case with [tex]p_1[/tex] and [tex]p_2[/tex]?

    Any help, on how best to estimate the paramters [tex]p_1[/tex], [tex]p_2[/tex] and [tex]y[/tex] would be great?
     
  2. jcsd
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook

Can you offer guidance or do you also need help?
Draft saved Draft deleted