5 ms·
> Can you show that [generating a number of records to skip instead of a yes/no for each record] will still maintain a uniform probability distribution? That s
by shoegumfoot 12y ago
> Can you show that [generating a number of records to skip instead of a yes/no for each record] will still maintain a uniform probability distribution?
That should work if you pull the skip count from the appropriate distribution (negative binomial, I think).
- eru 12y agoYes, you have to work out the right distribution, and how to sample from it (at all, and perhaps how to sample from it efficiently).