How to Generate random sampling from a probability distribution function generated by kernel density estimation?

Hi all,

I generate the PDF using ksdensity function matlab. Now I want to genrate random samples for the Monte Carlo simulation from that PDF, Anybody has an idea how to generate these random samples?

Many thanks in advance

MAnsour

Answer 1

[Bjorn Gustavsson]

Either you do something like this (Untested-unverified-off-the-cuff^TM):

% Assuming pdfKD values at x:
cdfKD = cumtrapz(x,pdfKD);
cdfKD = (cdfKD - cdfKD(1))/(cdfKD(end)-cdfKD(1)); % normalizing the cdf to be between zero and 1 at the end of your intervall
Q = rand(n);
randVals = interp1(cdfKD,x,Q,'pchip-or-linear');

Or you take a look at the file exchange contributions that solves just this task:

random-numbers-from-a-user-defined-piecewise-polynomial-distribution

arbrand-m

HTH