<from a terminating pass in Mathematical Statistics at university in 1978>
Determine the probability below which a data pile is not to be considered a Gaussian distribution.
For each of your data piles
Determine the mean and standard deviation of the relevant measurement in your data pile.
Plug that in to the Gaussian distribution formula to get your theoretical Gaussian distribution formula (null hypothesis).
Perform the Chi square test on the data pile and reject it (your null hypothesis) as not a Gaussian distribution if the probability is lower than that you determined.
The rest is left as an exercise for the reader (translation: I would actually have to look it up myself :) ).
From memory there are some results that say something like there should be no more than a certain percentage of the sample more than a certain number of standard deviations away from the mean. Calculating whether this is true of a sample may be enough. Again I can't remember the exact figures for this.
From: MIDRANGE-L [mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Booth Martin
Sent: Tuesday, 22 December 2015 10:31 AM
To: Midrange Systems Technical Discussion <midrange-l@xxxxxxxxxxxx>
Subject: Bell curves
I have a large data pile to review regularly and find exceptions. Its looking at first blush that I am looking at lots and lots of bell curves with a desire to isolate and review only the items that have irregular bell shapes.
This seems to me like a problem that must occur regularly in the DB2
world? There must already be paradigms for doing this? Any
experiences, pointers, or suggestions on ways to approach this?
Mirrors should reflect a little before throwing back images. -- Jean Cocteau
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a moment to review the archives at http://archive.midrange.com/midrange-l