26 Sep
2007
26 Sep
'07
10:09 a.m.
On Wed, 26 Sep 2007, apfelmus wrote:
ok wrote:
I believe the author may have misunderstood "numerically stable". The obvious (sum xs)/(fromIntegral $ length $ xs) is fine for the mean,
That's probably my fault, out of ignorance. Do you know a good online resource about numeric stability? (I don't have the Knuth at home. Didn't he say something about the mean formula? Or was it the standard derivation?).
"standard deviation"? I expect that indeed variance (thus also standard deviation) by incremental computation suffers from cancellation.