Mixture
This looks like a parameter mix distribution. Si a binomial distr with parameters (N,p) where the parameter N now is **not** fixed but varies as poisson with intensity say lambda. Then we have a new distribution which is poisson with intensity (lambda*p) [so mixture =a conditional distribution given the parameter and the distribution of the parameter itself ] I vaguely think this may be possible with the distrEx package. Best, Krishna
Spencer Graves wrote:
What problem are you trying to solve? Mixing Gaussian and hypergeometric distributions seems to me like mixing apples and volts: The Gaussian or normal distribution is continuous, while the hypergeometric is discrete. Poisson and binomial are both discrete, but the first represents counts of events assumed to be independent with no theoretical upper limit, while the binomial is the number of "successes" out of a fixed number of trials. This is more like comparing a rain storm with a bottle of water.