From koffice  Thu Feb 26 17:50:13 2004
From: Mark Arrasmith <arrasmith () math ! wichita ! edu>
Date: Thu, 26 Feb 2004 17:50:13 +0000
To: koffice
Subject: Statistical Tests ...
Message-Id: <200402261150.13713.arrasmith () math ! wichita ! edu>
X-MARC-Message: https://marc.info/?l=koffice&m=107781783209410

After reading this paper ...
	http://www.csdassn.org/software_reports.html

I decided to go through Kspread and check out its performance on statistical 
tests.  I've only gotten through the Poisson Distribution so far and KSpread 
1.3 seems to have the same problems as Excel.  

Also, the cell values are sometimes  correct, but the displayed contents are 
incorrect at times.  For example =POISSON(10,200,1) shows 0.00000E+00 in the 
cell, but the status bar sum shows the correct value of 4.10958 E-71.  But, 
the poisson algorithm fails for =POISSON(number,200,1) where number >= 134.

The author of the paper seems to have contacted the Gnumeric developers to 
help them with similar flaws.  I figure maybe a collaboration with author and 
the KSpread developers is in order?

- Mark Arrasmith

-- 
QOTD:
	"I've got one last thing to say before I go; give me back
	all of my stuff."
____________________________________
koffice mailing list
koffice@mail.kde.org
To unsubscribe please visit:
https://mail.kde.org/mailman/listinfo/koffice