I want to know how I can model a distribution that is basically n Bernoulli trials. However, these Bernoulli trials can have different probabilities and outcomes.
It is basically the same question as asked here: Probability distribution for different probabilities, with the addition of the possible different outcomes.
The eventual goal is to give a confidence interval on a measure of a set of n predicted revenue streams with it's own size and probability.
Is it even possible to define a confidence interval or define a distribution when the individual Bernoulli trials that make it up have different possible outcomes?