Check out reservoir sampling in wiki. On Sun, Jun 27, 2010 at 11:13, Anil C R <cr.a...@gmail.com> wrote:
> it is... but a small mistake in the code again... > def get_random_obj( objstream ): > count = 1 > obj = objstream.get() > while not objstream.end(): > * count += 1* > r = rand(0, count) > if r == 0: > obj = objstream.get() > else: > objstream.get() #discard > return obj > > note that you cannot make sense of equal probability if there are infinite > number of objects. Hence assume that there are N objects streaming. > > *Loop invariant*: probability of choosing any of the previously streamed > objects at the end of iteration n is > P[k, n] = probability that kth object is chosen at iteration n > *Base*: at the end of first iteration, P[1,1] = P[random(0,2)=0] = 1/2 > *Induction*: > Assume the invariant to be true at all iterations < n > > P[n, n] = P[random(0, n) = 0] = 1/n > P[k, n] = P[k, n-1]*(1-P[n,n]) = (1/(n-1))*(1 - (1/n)) = 1/n > > therefore at the end of N iterations, the probability of choosing any > object is 1/N > > Anil > > > > On Sun, Jun 27, 2010 at 9:19 AM, sharad kumar <sharad20073...@gmail.com>wrote: > >> @above >> probabilty for all numbers in d stream must be *same*.....whether first >> or last.... >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Algorithm Geeks" group. >> To post to this group, send email to algoge...@googlegroups.com. >> To unsubscribe from this group, send email to >> algogeeks+unsubscr...@googlegroups.com<algogeeks%2bunsubscr...@googlegroups.com> >> . >> For more options, visit this group at >> http://groups.google.com/group/algogeeks?hl=en. >> > > -- > You received this message because you are subscribed to the Google Groups > "Algorithm Geeks" group. > To post to this group, send email to algoge...@googlegroups.com. > To unsubscribe from this group, send email to > algogeeks+unsubscr...@googlegroups.com<algogeeks%2bunsubscr...@googlegroups.com> > . > For more options, visit this group at > http://groups.google.com/group/algogeeks?hl=en. > -- You received this message because you are subscribed to the Google Groups "Algorithm Geeks" group. To post to this group, send email to algoge...@googlegroups.com. To unsubscribe from this group, send email to algogeeks+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/algogeeks?hl=en.
