Re: Various kolmogorov-smirnov
From: Ross Clement (clemenr_at_wmin.ac.uk)
Date: 10/23/04
- Next message: Anon.: "Re: Randomized block design"
- Previous message: Top Spin: "Re: Averages, weighted, rolling, attenuated?"
- In reply to: Alex: "Various kolmogorov-smirnov"
- Messages sorted by: [ date ] [ thread ]
Date: 23 Oct 2004 00:48:45 -0700
Alex <hyp3r3xt3nsion@gmail.com> wrote in message news:<Itied.2088$%k.722@pd7tw2no>...
> Does anybody know the method that S-plus uses to calculate the kolmogorov-
> smirnov statistic ( i looked at the code of the function but it gave me a
> serious headache, heh).
>
> The reason I ask is because I'm trying to program it the way we were taught
> in school way back when (like this-> http://www.maths.qmw.ac.uk/
> ~bb/CTS_Chapter3_Students.pdf ), and I end up with different test
> statistics (the end result is the same, but the value of the test
> statistics is different and i can't figure out why).
Are you using the Kolmogorov-Smirnov statistic in the 'Stable'
library? (http://pws.prserv.net/jpnolan/SPlusUserManual.pdf). If so,
there's a not very detailed description in that pdf. Otherwise, if
you're using a different function, you could try the Stable library
and compare.
> Using the link i gave (where it gives 5numbers and tests whether they're
> from normal or not), doing it that way I get a test statistic of 0.326
> (like whoever did the pdf got). But doing it in splus, i get 0.2123. I
> can't figure out where the difference is coming from.
0.2123 is close to being twice the square root of 0.326. The square
root of 0.326 is 0.106276, and twice that is 0.212552. Calculated
using "bc -l".
Cheers,
Ross-c
- Next message: Anon.: "Re: Randomized block design"
- Previous message: Top Spin: "Re: Averages, weighted, rolling, attenuated?"
- In reply to: Alex: "Various kolmogorov-smirnov"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
