Mikey C wrote:
StdDev Variance
When implementing these, make sure *not* to use the "textbook" one-pass formula (mean of the squares - square of the mean); it simply doesn't work properly in floating point (you can find yourself subtracting one large number from another and losing most of your precision). I believe many versions of Excel made that mistake. There are some numerically stable one-pass algorithms for computing variance; search for them if you don't want to make two passes.
