Technical University Darmstadt, Germany
Department of Computer Science
Institute of Intellectics
Prof. W. Bibel


The Institute of Intellectics at the Technical University of
Darmstadt, Germany, offers two

		 =============================
		   PhD and PostDoc Positions
		 =============================

for our research projects LExIKON und DaMiT.


LExIKON is a research and development project focussing an innovative
approach to knowledge extraction from the Internet. The peculiarity of
LExIKON is to invoke inductive learning techniques on different
levels. By its way of integrating inductive inference into knowledge
extraction from the Internet, LExIKON is distinguished from all
competing projects throughout the world.  The LExIKON research deals
with the development of new learning algorithms, with their study and
with proofs of formal results about the algorithms' power and
limitations.  Those algorithms are prototypically implemented and
practically evaluated within the LExIKON project.  LExIKON puts
emphasis on the development of a core system named the Intelligent
Extraction Unit (IEU). Embedding LExIKON's IEU into a conventional
search engine may result into a breakthrough to the next generation of
those systems. The project partners are three german universities, the
German Research Center for Artificial Intelligence, two software
houses and a bank.

DaMiT aims at the development of a tutoring system for data mining.
Its pecularity is the fact that the topic and the location of teaching
coincide: extraction of information from huge and distributed data
bases as well as from the internet cannot be studied more realistic
than in the internet itselfes.  Within the projekt DaMiT, the
fundamentals of machine learning, ranging from inductive inference to
knowledge discovery and data mining, will be elaborated, will be
prepared for teaching in the internet and will be integrated in a
tutoring system.  Within this system, algorithms as well as complete
systems for data mining and knowledge discovery are accessible. Thus,
the students can probe their knowledge practically.  The project
consortium consists of 10 german universities.


keywords: artificial intelligence, machine learning, knowledge
extraction, data mining
start: promptly possible
project descriptions: http://lexikon.dfki.de http://damit.dfki.de
contact and further information: Gunter Grieser, Tel. +49+6151+166634,
grieser@...

I'm a postdoc in Pfizer's bioinformatics technology team.


I'm interested in pattern searching on biological sequences, unsupervised
classifications and clustering of this data, and more.

I've got a background in physics, mathematics, molecular biology, and
computation.

I'm just getting into the mathematical aspects of  pattern recognition. I
feel like I'm swimming in a Bayseian prior sea. :)

Thanks for this list!
Deanne

I've created a new marketing research discussion list named
"marketing-research@yahoogroups.com". The emphasis of the new list is
on the applications of modern modeling methodology and techniques from
Statistics, Data Mining, and Machine Learning in Marketing
Research. The discussion list would serve as a forum where marketing
researchers, marketing professionals, statisticians, data miners, and
data analysts can meet, discuss modeling issues, and exchange ideas.

Details of the list can be found at

    http://groups.yahoo.com/group/marketing-research

Thank you for your attention.



--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com



------------------------------------------------------------
Get paid to write review! http://recursive-partitioning.epinions.com

Please be advised the March 2001 issue of SecondMoment has been
posted at http://www.secondmoment.org .  This month we are very
pleased to feature an article on causality and Bayesian Networks
written for SecondMoment by Dr. Judea Pearl. Dr. Pearl is an Emeritus
Professor at University of California, Los Angeles, Computer Science
Department.

As always we welcome your comments, and invite you to submit any
papers or articles you would like to share with our readers.

Yong Kim
Site Editor
SecondMoment.org

I've listed some patents on decision trees awarded by the US Patent and
Trademark Office at

    http://www.recursive-partitioning.com

under Bibliography/Patents.



--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com



------------------------------------------------------------
Get paid to write review! http://recursive-partitioning.epinions.com

Please note the December issue of SecondMoment has been posted at
http://www.secondmoment.org . This month we are featuring an article
on statistics and web traffic analysis. As always we welcome your
comments, and invite you to submit any papers or articles you would
like to share with our readers.

The following is a message I received from Jay Magidson regarding the history of
CHAID.


>From: "Jay Magidson" <jay@...>
>To: <tslim@...>
>Cc: "Babinec, Tony" <tbabinec@...>
>Subject: chaid
>Date: Wed, 22 Nov 2000 11:25:52 -0500
>
>Date: Nov. 22
>To: T.S. Lim
>From: Jay Magidson
>
>Gordon Kass developed CHAID as part of his dissertation at U of
>Witwatersrand, South Africa.  He published in the journal Applied Statistics
>and offered to send his PL1 code to anyone upon request.  Bill Perreault at
>U of North Carolina requested a copy and had a graduate student add some
>user friendly features and distributed that PL1 version for a few years.
>
>I, through my company Statistical Innovations (SI), developed a Fortran
>CHAID program and called it SI-CHAID.  Originally, it was licensed on PCs in
>DOS, and later WINDOWS 3.1.  I added several features, including a new
>algorithm for ordinal dependent variables (Kass' original CHAID algorithm is
>designed only for nominal dependent vars).  I have authored several
>publications on CHAID.  I sold the WINDOWS version to SPSS and they
>incorporated it into ANSWER TREE.  I also developed 2 SAS PROCs called
>SI-CHAID and SI-DRAW (for IBM mainframes running under MVS -- now called
>OS3 -- which are marketed through SI.  Gordon Kass is a licensee of my
>SI-CHAID SAS PROC.
>
>I hope this is helpful.  Feel free to request any additional info, and
>please send me a copy of any info that you publish on CHAID.
>
>Thank you




--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com



------------------------------------------------------------
Get paid to write review! http://recursive-partitioning.epinions.com

I was wondering if anyone here knows the history of CHAID. Who was the first
person who programmed the algorithm as described in Kass (1980)? Did Kass ever
program it himself? Thanks.



--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com



------------------------------------------------------------
Get paid to write review! http://recursive-partitioning.epinions.com

"If you don't mind, please send out a short introduction about
yourself and your interests in tree-structured methods to the list.
Thank you."

I'm a computer programmer working at Duke University Center for
Demographic Studies interested in the methodology of recursive
partitioning as applied to analysis and interpretation of multiple
complex pathways to many illnesses, diseases, and the bio-demography
of aging.

rfp

SAS has teamed with e-Intelligence expert Ismail Parsa of Epsilon to offer a
three-day seminar entitled Mining Web Data for Personalization Solutions. (Full
details at http://www.sas.com/bks/webmining).

DECEMBER 6-8 in BOSTON, Massachusetts, is the next scheduled seminar.

From this seminar, you will gain an understanding of mining consumer
navigational patterns and purchase behavior to support marketing, sales and
personalization solutions as they relate to one-on-one targeted marketing.

Topics addressed are:
      * Data Sources on the Web, Enhancing Web Data, and Integration with Offline
Data
      * e-Metrics: Measuring Success on the Web
      * Clickstream Analysis for Personalization
      * Overview of Web Log Analysis, Web Mining Solutions/Technologies, and
e-Business Platforms

Ismail Parsa will be presenting the seminar at these locations:
      Boston, Massachusetts .........Dec. 6-8, 2000
      Orlando, Florida..................Jan. 24-26, 2001
      Phoenix, Arizona.................Feb. 14-16, 2001

To learn more about the seminar visit: http://www.sas.com/bks/webmining

~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Judy Davis
SAS Institute Inc.
Training Division Marketing
(919) 531-8495
Judy.Davis@...

The Power to Know(tm)

SAS has teamed with CRM experts Michael Berry and Gordon Linoff of Data Miners
to offer a three-day seminar entitled Customer Relationship Management Through
Data Mining. (Full details at http://www.sas.com/bks/crm)

DECEMBER 18-20 in HOUSTON, Texas, is the next U.S. scheduled seminar, and
Michael Berry will be the instructor.

From this seminar, you will learn how to apply data mining at the critical
points of the customer lifecycle:
      * Acquisition
      * Retention
      * Loyalty

You will also be lead through a process that will help you achieve vital CRM
goals:
      * Increase customer lifetime value
      * Segmentation of customers by profitability
      * Segmentation of customer by risk of default/churn
      * Integration of data mining into the full marketing process
      * Improved customer retention
      * Increased wallet share

Here's the upcoming schedule:
      Munich, Germany...............Dec. 1, 2000 (one-day seminar) - Michael
Berry instructing
      Amsterdam, Netherlands.......Dec. 18-20, 2000 - Gordon Linoff instructing
      Houston, Texas..................Dec. 18-20, 2000 - Michael Berry
instructing
      San Francisco, California......Jan. 17-19, 2001 - Gordon Linoff instructing

To learn more about the seminar visit: http://www.sas.com/bks/crm

~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Judy Davis
SAS Institute Inc.
Training Division Marketing
(919) 531-8495
Judy.Davis@...

The Power to Know(tm)

University of Aberdeen

Department of Engineering

Research Studentship available:

Neural systems engineering for embedded machine vision

Applications are invited for a fully-funded, three-year PhD
studentship.  The work will be researching future artificial neural
network (ANN) architectures based on our neural hardware.

The types of vision application which will form the basis for mass
implementation over the next few years are those requiring the
embedding
of intelligent pattern recognition systems into industrial processes,
instrumentation and portable systems.  These systems will make extreme
demands on the hardware, requiring the integration of traditional
image
processing with ANN techniques, and implemented in designs with small
footprint and low power consumption.  The project will research
methods
for the efficient embedding of imaging algorithms into hardware and
neural arrays.  A hardware implementation of the system will be
developed.
  The work will be motivated by industrially relevant demonstrator
applications, and will include collaboration with system developers
and
end users.

The studentship is partly funded by AXEON Ltd who are developing novel
processor systems.  The student would join a lively and rapidly
expanding novel architectures research group in the Department of
Engineering.
http://www.eng.abdn.ac.uk/dept/research.html

The funding of the studentship covers UK/EU tuition fees and a
maintenance grant equivalent to standard Research Council rates.
Applicants should have a good first degree (equivalent of UK 1st class
or upper second) in electronic engineering, physics or computational
science.  Informal enquiries may be directed to Dr Alastair Allen.
Email: a.allen@....  Tel: +44 1224 272501.  Fax: +44 1224
272497.

Application forms: Postgraduate Office, Department of Engineering,
Fraser Noble Building,  University of Aberdeen, Aberdeen, AB24 3UE,
UK.
Tel: +44 1224 272513.  Fax: +44 1224 273895.  Email:
pgoffice@...

Please note the November issue of SecondMoment has been posted at
http://www.secondmoment.org . This month we are featuring a
discussion with Dr. Hal White on the subject of Reality Check for
data mining, a follow-up to last month's discussion on artificial
neural networks. As always we welcome your comments, and invite you
to submit any papers or articles you would like to share with our
readers.

Yong Kim
SecondMoment

T.J. Lim wrote:
> At least, the algorithms are described openly. You just need to be
> patient to disect and compare them.

True, but what if they weren't and in both cases they were proprietary?  And
you just got wildly differing results?  For myself, I just go with stuff
that has been spelled out in detail and not 'black boxed', so I can see what
happens.

>
> It's their lost if they avoid Statistics though. :(

I agree, but right now the whole field is in upheveal intellectually and the
debate over the book DARKNESS IN EL DORADO isn't helping it.

Best, Mark Hall

T.S. Lim said
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
Since most (if not all) applications of Data Mining
are in commerce, the risk of using unproven methodology that hasn't
been extensively scrutinized may be acceptable.
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

I once asked Gwilym Jenkins, of Box-Jenkins forecasting fame, why
he was willing to sell commercial forecasting software with trade
secrets embedded when everyone knew that scientific peer review
was the only way to ensure that we were doing good science.

He replied that acceptance and use in the commercial world was
an alternative community to scientific peers for setting a
standard of excellence.

My opinion is that he was right: if you are willing for your
company to make bad decisions, you too can help prove that
the unreviewed science behind software is invalid.

George S. Cowan

>
> When dealing with proprietary methodology, it's (practically)
> impossible to study the properties of the method thoroughly.
Personally, I
feel uncomfortable using a method that can't
> be evaluated objectively by fellow researchers. It may be OK if the
> application has nothing to do with human experimentation (as in
> Biostatistics). Since most (if not all) applications of Data Mining
> are in commerce, the risk of using unproven methodology that hasn't
> been extensively scrutinized may be acceptable.

You aren't the only one.  As an archaeologist who extensively uses
statistics,
it really helps to know the assumptions and how the algorithm is
working.
As an example, Mike Baxter from the Nottingham-Trent University has
some
forthcoming papers (currently available by snail mail as departmental
technical reports) on the use of model-based clustering in analyzing
geochemical data from pottery.  Very different results occur
depending on
the program one uses (in this case EMMIX and MCLUST) due to the
implementation of the EM algorithm and the maximization criteria.

So, combine that with the fact a lot of our data is non-normal to
begin
with, and its no wonder most archaeologists these days avoid anything
with
math, statistics, etc.

Sorry these comments have little to do with RP per se, but it is a
very
similiar situation though.

Later, Mark Hall

In the Data Mining world that is dominated by Computer Scientists, the
methodology behind the software packages sold/licensed in the market
is often proprietary. Take, for example, the classification and
regression trees software package CART(r). The basic idea behind
CART(r) is the algorithm proposed by Breiman, Friedman, Olshen, and
Stone (1984). However, there has been quite a few proprietary
improvement in CART(r) so that you can no longer know for sure what's
going on inside the software package. The same is true for C5.0/See5
(another classification trees software) that supersedes C4.5.

When dealing with proprietary methodology, it's (practically)
impossible to study the properties of the method thoroughly. Personally, I feel
uncomfortable using a method that can't
be evaluated objectively by fellow researchers. It may be OK if the
application has nothing to do with human experimentation (as in
Biostatistics). Since most (if not all) applications of Data Mining
are in commerce, the risk of using unproven methodology that hasn't
been extensively scrutinized may be acceptable.

Perhaps this joke is true after all: when a Statistician gets an idea,
she/he'll write and publish a paper while when a Computer Scientist
gets an idea, she/he'll form a company. :)

Comments?



--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com



------------------------------------------------------------
Get paid to write review! http://recursive-partitioning.epinions.com

Once again I would like to invite you to visit Second Moment at
http://www.secondmoment.org , a new website for applied statistics
and analytics. This month we are featuring our interview with Dr.
Halbert White on artificial neural networks. Dr. White is a professor
of Economics at University of California, San Diego and is one of the
foremost experts on the subject. Also if you have an article or a
paper that you would like to share with our readers, please feel free
to submit it on our site. We welcome and appreciate your
participation.

Yong Kim
Second Moment

Once again I would like to invite you to visit Second Moment
(http://www.secondmoment.org), a new website for applied statistics
and analytics. This month we are featuring an article titled: Load
and Price Forecasting Using Statistical Methods and Models. The
article was submitted to us by Dr. Michael Smith, a senior lecturer
at the University of Sidney, Australia in the department of
Econometrics and Business Statistics. As always, we welcome your
comments. Thank you.

Yong Kim
Second Moment

I want to thank everyone who visited and provided feedback on Second
Moment (http://www.secondmoment.org), the new Web site for applied
statistics and analytics. The response has been extremely positive so
far, and it appears that Second Moment is on the right track in its
mission to bring together the interests of academia and industry.
Please note this month's feature article focuses on cluster
analysis. In addition, we have added a message board to promote
discussion among our readers. Once again we welcome your visit and
comments. Thank you.

Yong Kim
Site Editor

I am hoping to get some feedback regarding an applied statistics and
analytics website I created to bring together academics and industry
analysts. It is designed to showcase academic research and create
discussion in the field of applied statistics and to provide links to
other useful statistics sites as well as provide opportunities for
interesting work. The site is called Second Moment and
is at <http://www.secondmoment.org>. I would appreciate your feedback
and suggestions on how I could improve the site. Also, if
appropriate, please fee free to add a link to any of your sites.
Thanks.

I've come across the following new book on classification and regression trees:

Graphes d'induction
by Ricco Rakotomalala and Djamel Zighed
Hermes Science Publications, Paris, France
2000

If someone in the lists speaks French and happens to have read the book,
I'd be interested in hearing your opinion about the book. Thanks.

--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com
______________________________________________________________________
Get paid to write a review! http://recursive-partitioning.epinions.com

The following web page criticizes decision trees and contrasts it to
cluster analysis.

        http://www.clustan.com/clustering_v_decision_trees.html

I just wonder how you can compare cluster analysis (which is unsupervised
learning; i.e., there's no response variable) to decision
trees (which is supervised learning; i.e.,  there exists a response variable).

--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com
______________________________________________________________________
Get paid to write a review! http://recursive-partitioning.epinions.com

To celebrate the first anniversary of Recursive-Partitioning.com (May 4,
2000), I've redesigned the website. It's now faster to load
and much easier to navigate. You can rate and even review a resource. It's
also easier to suggest a site to be added. Thank you for your attention.

--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com
______________________________________________________________________
Get paid to write a review! http://recursive-partitioning.epinions.com

I'd like to announce Knowledge Discovery Central (Preview), my attempt at
creating a new Web portal dedicated to knowledge and information discovery
from data. I'm collecting relevant links from Statistics, Biostatistics,
Data Mining, Machine Learning, Knowledge Discovery in Databases,
Economics, Sociology, etc. There're currently 456 links in the database.

I welcome your contributions. The URL is

    http://www.kdcentral.com

Note if you'd like to submit a personal Web page, it must contain a unique
and original content.

Thank you for your attention.

--
T.S. Lim
tslim@...
www.Recursive-Partitioning.com
______________________________________________________________________
Get paid to write a review! http://recursive-partitioning.epinions.com

-----------------------------------------------------------------------

      Postdoctoral Fellowship in Biostatistics / Bioinformatics
      Division of Biostatistics, Oncology Center
      Johns Hopkins University

------------------------------------------------------------------------

The Division of Biostatistics in the Oncology Center of the Johns
Hopkins University is inviting applications for one position of
postdoctoral fellow to work in the area of biostatistics and
bioinformatics for gene expression array analysis. A suitable applicant
will hold a PhD in computer science, statistics, or
biostatistics. Candidates with a PhD in areas genetics or microbiology,
and substantial quantitative expertise, are also encouraged to
apply. Candidates should have evident potential for excellence in
interdisciplinary quantitative research, and interest in becoming
familiar with issues in genomics. Some background in one or more areas
such as clustering, classification, discriminant analysis, model
selection, or Bayes nets is desirable. The position is available
immediately and for up to two years.

Applicants should mail or e-mail cv (including a personal website url if
available) and letter of application, and arrange for two or more
letters of reference to be sent to: Postdoctoral Search Committee,
Division of Biostatistics, 550 North Broadway, suite 409 Baltimore, MD,
21205-2011. Applications will be considered until the position is
filled. E-mail inquiries on the status of an application can be
directed to mooreal@.... Johns Hopkins University is an Affirmative
Action/Equal Opportunity Employer.

------------------------------------------------------------------------

try posting your question at the lists class-l and mpsych-l

____________________Reply Separator____________________
Subject:    [RP] question: RFCM
Author: Svitlana Bulashevska <s.bulashevska@...>
Date:       02/28/2000 10:29 AM

Could anybody give some references on relational fuzzy clustering
method (RFCM)?
May be attach some papers?
Thank you very much.




------------------------------------------------------------------------
To UNSUBSCRIBE,  send an empty message to

recursive-partitioning-unsubscribe@egroups.com

To SUBSCRIBE,  send an empty message to

recursive-partitioning-subscribe@egroups.com

Archives: http://www.egroups.com/group/recursive-partitioning

------------------------------------------------------------------------
Get what you deserve with NextCard Visa! Rates as low as 2.9%
Intro or 9.9% Fixed APR, online balance transfers, Rewards Points,
no hidden fees, and much more! Get NextCard today and get the
credit you deserve! Apply now! Get your NextCard Visa at:
http://click.egroups.com/1/949/5/_/8970/_/951730195/

-- Talk to your group with your own voice!
-- http://www.egroups.com/VoiceChatPage?listName=recursive-partitioning&m=1

Could anybody give some references on relational fuzzy clustering
method (RFCM)?
May be attach some papers?
Thank you very much.

Dear all,

I am looking for a good book recommendation concerning classification
trees.

So, if you have a good book, please send my your response.

Best wishes

Wolfgang

gsbin-@... wrote:
original article:http://www.egroups.com/group/recursive-partitioning/?s
tart=64
> We are trying to locate papers that describe the methods used in
Rpart for
> n-fold cross-validation for regression trees.  Any suggestions?
> Sincerely,
> Greg Binns

Have you looked at the 2 tech. reports that come with the RPART
distribution? The N-fold cross-validation for regression trees is the
same as that for classification trees (or Poisson regression/survival
trees).