JOINT HALF DAT MEETING BETWEEN THE RSS STATISTICAL COMPUTING SECTION AND
ASSUME
Wednesday 17th March 2004, 2:00-5:30 pm
Errol Street
Tea will be served at 3:45pm.
Enquiries
Phil Bowtell (+44 (0)1784 498374)
[log in to unmask]
Deatils given below and on our website
http://www.gla.ac.uk/External/RSS/RSScomp/spr3mtg.html
This joint half day session with ASSUME is split into two parts. The first
sessions discusses statistical 'add-ins' and the importing of R and NAG
routines into Excel to allow more complex and powerful statistical
calculations. After tea the concept of building array calculations in
Excel is discussed.
"Statistics on the desktop: Integration of Excel and R"
Erich Neuwirth (University of Vienna )
Synopsis: Microsoft Excel is probably the most widely used tool for
statistics. On the other hand, is has not been designed as a statistical
tool, therefore it does not offer all the advanced statistical methods
professional statisticians need. We will show how R, a powerful open
source statistics program can be integrated into Excel and how this
integrated environment allows to combine the drag and drop interface and
the spreadsheet concepts with the powerful statistical system offering
facilities to develop very advanced statistical procedures. We will also
discuss how different user groups can benefit from this combination in
different ways. Some examples will show how this software environment can
be used to rapidly develop powerful interactive statistics applications.
"Importing NAG Routines into Excel"
David Sayers (NAg Ltd., Oxford, UK )
Synopsis: This talk will illustrate the ease with which routines written
in Fortran or C may be incorporated into Excel without the need to
translate the source into VBA. Throughout the presentation copious
examples will be given, starting first with a peek at the insides of the
NAG Statistical Add-Ins, followed by a series of examples linking other
NAG routines to Excel. It is assumed throughout that users may have
routines of their own that they would like to use within the Excel
environment and this talk indicates how this might be achieved.
Principally the Add-In route is illustrated, but an example of a COM
interface is also given. Rather than concentrate on details, an overview
is given but the author will be pleased to receive particular questions
either during the meeting or separately offline.
Francois Sermier (Independent Consultant, Paris )
Synopsis: Excel has few data types (number and text mainly) but, since
the beginning, among them there is an array type. So many know that there
are TRANSPOSE, MMULT or MINV functions to achieve mathematical array
calculations. As an example of such calculations, we will show weighted
linear regression.
Many users (and surely all statisticians) know of "array functions", the
ones that need to be entered with "Ctrl-Shift-Enter". As a minimum, they
know the FREQUENCY function to calculate a (one-way) distribution of
quantitative data.
But Excel (the worksheet) may also calculate algebraic operations
*directly* on array and moreover *apply* worksheet functions to array.
This means for example that LOG(list) -> list of LOG of each value in the
list. Unfortunately not all functions are prone to this kind of
calculation.
Knowing this, it is possible to use the worksheet as a sort of "functional
programming" environment, a little APL-like.
I will show how to use it, how it may change the vision of data and
calculations in Excel and give some tricks. The last example will be a
little "tour de force": calculation of the independance chi-2 statistics
from a two-way contingency table in one (and only one !) cell.
|