Exact relative error distribution for estimating coin probabilities

09/10/2013 - 12:15pm
09/10/2013 - 1:10pm
Mark Huber (CMC)

Suppose that I have a coin with unknown probability $p$ of heads. One way to estimate $p$ is to flip the coin $k$ times, count the number of heads, and divide by $k$. While this is in some sense the more efficient method, the relative error in this approach depends on $p$, the unknown quantity. In this talk I will show how to build $\hat p$, an estimate for $p$, such that the relative error in the estimate, $\hat p/p - 1$, is completely independent of $p$. Along the way we'll use an interesting fact about generating functions for random sums of random variables.

Mudd Science Library 126, Pomona College
Misc. Information: 

There will be a short organizational meeting preceding the talk at 12:00 noon.

Claremont Graduate University | Claremont McKenna | Harvey Mudd | Pitzer | Pomona | Scripps
Proudly Serving Math Community at the Claremont Colleges Since 2007
Copyright © 2018 Claremont Center for the Mathematical Sciences