# Predicting time interval based on statistics

Let's say, we have some random event. We also have a hist of time intervals between two events, based on statistical data. For example, a frequency distribution:

[100,30,15,10,7,6,5,5,4,4]


Each bin value corresponds to length of time interval between two events. 1st bin corresponds to [0,1) seconds, 2nd bin is for [1,2) seconds, 3rd is for [2,3) secs and so on.

How would I predict the time that it might take to get K events?

If we normalize and cumulatively add the bin's value, we can get a probability hist, each bin's value is P(one event in time < bin's edge). But how would I get such a probability hist for K events? P(K events in time < bin's edge)

-
Anyway, read this first en.wikipedia.org/wiki/Poisson_process –  belisarius Oct 8 '12 at 18:09
Google for negative binomial distribution. In addition to Wikipedia, one good site is math.uah.edu/stat. –  user21171 Feb 22 at 23:28