- #1
omg!
- 52
- 0
hi all,
the situation is this:
i have a conditional PDF [tex]f_{X|(D,S)}(x|d,s)[/tex] and the unconditional PDFs D and S, [tex]f_{D}[/tex], [tex]f_{S}[/tex].
Now, I can marginalize the joint PDF [tex]f_{X,D,S}=f_{X|(D,S)}f_{D}f_{S}[/tex] by integrating over [tex]d[/tex] and [tex]s[/tex] over their respective supports. What I'm really interested in is a distribution parameter of the random variable D, let's call it [tex]p[/tex]. The marginal PDF f_{X}(x|p) is the entity at interest, and for example I could perform MLE to determine the parameter [tex]p[/tex].
My question is as follows: Do you know a method that can do without explicitly calculating the marginal PDF [tex]f_{X}[/tex], i.e. not requiring integration which is computationally slow.
I really do not have a good overview over all the methods in statistics. From the sources that I was able to find, the methods Expectation-maximization algorithm, MCMC method were mentioned frequently but I don't understand how they work and how they can be useful for my problem.
thanks a lot!
the situation is this:
i have a conditional PDF [tex]f_{X|(D,S)}(x|d,s)[/tex] and the unconditional PDFs D and S, [tex]f_{D}[/tex], [tex]f_{S}[/tex].
Now, I can marginalize the joint PDF [tex]f_{X,D,S}=f_{X|(D,S)}f_{D}f_{S}[/tex] by integrating over [tex]d[/tex] and [tex]s[/tex] over their respective supports. What I'm really interested in is a distribution parameter of the random variable D, let's call it [tex]p[/tex]. The marginal PDF f_{X}(x|p) is the entity at interest, and for example I could perform MLE to determine the parameter [tex]p[/tex].
My question is as follows: Do you know a method that can do without explicitly calculating the marginal PDF [tex]f_{X}[/tex], i.e. not requiring integration which is computationally slow.
I really do not have a good overview over all the methods in statistics. From the sources that I was able to find, the methods Expectation-maximization algorithm, MCMC method were mentioned frequently but I don't understand how they work and how they can be useful for my problem.
thanks a lot!