what is the best probability distribution for modeling these situations?
easy version:
> a random integer between 1 ond n is generated at a given frequency
> the amount of possible integers is large and follows a perfect normal distribution
> what distribution will tell us how many periods will probably go by before we reach a cumulative sum larger than X?
hard version:
> same as above
> only difference is that the returned integer occasionaly could be zero/error/missing
> the frequency of zeros is unknown but non-random and changes over time
> if the returned integer is non-zero, the numbers follows the usual normal distribution
> what distribution will tell us how many periods will probably go by before we reach a cumulative sum larger than X?
the second version would be impossible without knowing at least some history, but let's say that these situations could be monitored over time, so regressions and such is possible
exact answers for X are for amateurs, good models give answers like "X will probably have happened with a probability larger than .50/.90/.95/.99/(choose given probability) at Y frequencies"
easy version:
> a random integer between 1 ond n is generated at a given frequency
> the amount of possible integers is large and follows a perfect normal distribution
> what distribution will tell us how many periods will probably go by before we reach a cumulative sum larger than X?
hard version:
> same as above
> only difference is that the returned integer occasionaly could be zero/error/missing
> the frequency of zeros is unknown but non-random and changes over time
> if the returned integer is non-zero, the numbers follows the usual normal distribution
> what distribution will tell us how many periods will probably go by before we reach a cumulative sum larger than X?
the second version would be impossible without knowing at least some history, but let's say that these situations could be monitored over time, so regressions and such is possible
exact answers for X are for amateurs, good models give answers like "X will probably have happened with a probability larger than .50/.90/.95/.99/(choose given probability) at Y frequencies"
