Lognormal distribution
- From: Carl Betterton <carlb uga edu>
- To: Morten Welinder <mortenw gnome org>
- Cc: Gnumeric Mailing List <gnumeric-list gnome org>, "Andreas J. Guelzow" <aguelzow pyrshep ca>, Jean Bréfort <jean brefort normalesup org>
- Subject: Lognormal distribution
- Date: Mon, 01 Sep 2008 13:59:21 -0400
I agree, and believe the Gnumeric function RANDLOGNORM does exactly what
the Excel function LOGINV does, if a random probability is used in the
Excel function. The Excel help documentation explains the syntax of
LOGINV(probability,mean,standard_dev) and gives an example. If one uses
the Excel function LOGINV(RAND(),mean,standard_dev) the result is the
same as with Gnumeric's RANDLOGNORM function. Actually, looking at the
Excel help file was how I came to properly interpret the Gnumeric function.
My original point, perhaps poorly made, was that the help documentation
for Gnumeric was not very helpful. I agree using semantics that are
consistent with Excel is a good approach, but the Gnumeric help
documentation should stand on its own should it not?.
Morten is spot on that R gives better consistency. If it were only
myself, R fills the ticket, but one of my goals is to have my students
accept Gnumeric as a viable (or preferable) alternate to Excel. Thus my
concern with help documentation.
As mentioned in the earlier post, I am new to Gnumeric and this list, so
am traveling the learning curve. Thanks.
Carl
Morten Welinder wrote:
It would make sense to first verify that LOGNORMAL does what it is
supposed to do, i.e., what it does for Excel. It is well known that
arguments to Excel's statistical functions can be illogical or worse.
Nevertheless, changing the semantics away from Excel's is only going
to make things worse.
If you need consistency, use the "R" functions: R.DLNORM, R.PLNORM,
and R.QLNORM.
Morten
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]