I'm writing a programm in a programming language which has a function to generate uniform distributed random values.
I need to make "asymmetric normal distributed" values out of them. By that I mean that it should be more likely to get values lower than the median than higher ones. I know that this is probably not called a normal distribution anymore, but I wasn't able to come up with a better explanation.
I have two more annotations:
- I don't want to get the median itself, so it should have a probabilty of 0%. I could just execute the algorythm again if it wouldn't accomplish this, so the point isn't that important
- If it makes it easier I only need a binomial distribution, because I need integer values in the end. But my researches so far showed more results using a normal distribution and rounding afterwards.
I already found out how to do this with a typical normal distribution without the asynchronous aspect here on math.stackexchange: http://math.stackexchange.com/q/69245/12864
And I came up with the idea that maybe two compound normal distributions would do the trick:
