> I only added the accuracy of the SD-algorithm into the discussion,
> because the naive single pass algorithm used inside the ztest routine
> may give very wrong results,
I'm sure you are right - but perhaps this should be a separate
issue that can indeed wait for a while?
This separate problem seems to be that ZTEST calculates variance
for itself, using the old formula that VAR originally used:
s^2 = (1 / (N - 1)) * [Sum1..N(Xi^2) - {(Sum1..N(Xi))^2 / N}]
In VAR, this formula/algorithm was replaced/improved in 2004
under issue 22811, because in extremis it falls over. From what I
can gather there are very few occasions when it matters - but of
course it still should be made as good as possible.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
