Re: Optimisation of statistic calculations

[Ben Pfaff]

John Darrington <address@hidden> writes:

> Like I said, there are ways to calculate variance that don't need any
> pre_dependencies, but there are other stats for which there isn't.
> The Levene statistic is one of them.  Thus, you will notice that
> levene.c makes two passes through the data.
>
> Does this make sense?

Yes, thanks for the explanation.  We really have two kinds of
dependencies that need to be handled differently.  Interesting, I
hadn't thought about that problem before.  I look forward to
seeing your implementation.

By the way, I see that Jason Stover has joined this discussion.
I don't know if Jason has been introduced here.  He has been
working on GSL for some time now and several months back
expressed interest to me about working on PSPP.  Jason's
expertise is in, as I understand it, numerical and statistical
code.  I don't think he's decided yet what he'd like to work on
in PSPP, but we can expect it to be in that realm.
-- 
"...In the UNIX world, people tend to interpret `non-technical user'
 as meaning someone who's only ever written one device driver."
--Daniel Pead