(text)
DG, EUFIT '98, Aachen, Germany, Sep 7-10, 1998, http://www.mitgmbh.de/elite/eufit.html
--
Data Mining and Knowledge Discovery community, focusing on the
latest research and applications.
Submissions are most welcome and should be emailed, with a
DESCRIPTIVE subject line (and a URL) to gps.
Please keep CFP and meetings announcements short and provide
a URL for details.
KD Nuggets frequency is 2-3 times a month.
Back issues of KD Nuggets, a catalog of data mining tools
('Siftware'), pointers to Data Mining Companies, Relevant Websites,
Meetings, and more is available at Knowledge Discovery Mine site
at http://www.kdnuggets.com/
********************* Official disclaimer ***************************
All opinions expressed herein are those of the contributors and not
necessarily of their respective employers (or of KD Nuggets)
*********************************************************************
~~~~~~~~~~~~ Quotable Quote ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
'If I had six hours to chop down a tree,
I'd spend the first four sharpening the axe'
- Abraham Lincoln (thanks to Stephen Koo) Previous1NextTop
Date: Mon, 5 Jan 1998
From: GPS gps
Subject: What Wal-Mart might do with Barbie association rules
Here is a summary of answers to a question asked by
Charles P. Elkan elkan@cs.columbia.edu
in KDNuggets 97:35,
who wrote:
Wal-Mart knows that customers who buy Barbie dolls (it sells one
every 20 seconds) have a 60% likelihood of buying one of three types of
candy bars. What does Wal-Mart do with information like that? 'I don't
have a clue,' says Wal-Mart's chief of merchandising, Lee Scott.
Source: Palmeri, Christopher. Believe in yourself, believe in the merchandise.
Forbes v160, n5 (Sep 8, 1997):118-124.
The best answer was given by Doron Shalvi doron@eng.umd.edu,
who suggested the mentioned candies should be manufactured in
the shape of a Barbie doll!
Most other suggestions were to put the Barbie and the candy closer together,
or packaging Barbie, candy and perhaps other products together,
or offering 'affinity program' that give Barbi
accessories in exchange for proofs of purchase.
Here are suggestions from other readers (edited for space):
By increasing the price of Barbie doll and giving the type of candy bar free,
wal-mart can reinforce the buying habits of that particular types of buyer.
Once this is done, the next time a buyer buys something else along with
this combination, those items from name-brand substitute manufacturer can
be placed suitably in the next aisle. A reorganization of the aisle content
will happen which which differ geographically. Users can buy more stuff
spending less time, increasing sales (and credit debt, of course)
Quantity discounts can be obtained from the manufacturer of that candy since
some amount of sales along with barbie is always assured.
My idea (probably pretty obvious) would be that retail outlets might use
the association rules (in general) for choosing what kind of products
they put close to each other. Now they probably won't put the candy bars
next to the Barbie dolls, but then they might introduce a sort of cross
reference signs: e.g. a sign above the Barbie dolls stating where the
candy bars are (or the other way around).
(1) Highest margin candy to be placed near dolls.
(2) Special promotions for Barbie dolls with candy at a slightly higher
margin.
(3) Coupons for dolls and candy at different times or places.
(4) Package 2 or more candy bars with Barbie dolls.
It seems likely that there will be a high correlation between
any customer purchases and whatever is placed for sale
at the checkout (i.e. candy) as an impulse buy. In these circumstances
Wal-mart should attempt to maximise its profits by placing higher
profit-margin goods near the checkout and during a specific
period discover the maximum expenditure/profit ratio customers will
devote to an impulse purchase.
Otherwise, if a strong
correlation truly exists between two products, Wal-mart can:
- exploit discovered associations with the companies who
manufacture the products with tie-ins
- create buy one, get one type offers (.e.g. buy a candy multipack for
a free Barbie hairbrush!) to increase sales based on this association
- Take a poorly selling product X and incorporate an offer on this which
is based on buying Barbie and Candy. If the customer is likely to buy
these two products anyway then why not try to increase sales on X?
Probably they can not only bundle candy of type A with Barbie dolls,
but can also introduce new candy of Type N in this bundle while
offering discount on whole bundle. As bundle is going to sell because of
Barbie dolls & candy of type A, candy of type N can get free ride to
customers houses. And with the fact that you like something , if you see
it often, Candy of type N can become popular.
Also they can try to increase the sell by hiking price of candy of type
A while lowering price of barbie dolls. Thus cut in dolls price can be
taken by hike in candy of type A & thus getting more money for pair
together.
As they introduce new & new candy's in this bundle, they can lower dolls
price
& keep hiking candy's price. Now you have many candy's in this bundle, so
little hike in candy's price will mint you a lot money.
Now can you guess which software empire build his company with this idea
in mind?
It's true that association rules and unsupervised learning can have
good applications in business, but not every discovery is a gold
nugget. Nevertheless, this one might have some practical
implications.
I wouldn't package candy bars with Barbi dolls -- the dolls may stay
on the shelf longer than the life of the candy. In addition, the
dolls are manufactured off-shore, while the candy is made here,
requiring extra steps for 'in-packing.'
Instead, how about an 'affinity program' that offers Barbi
accessories in exchange for proofs of purchase. This rewards
consumers for maintaining the association and benefits both
companies.
A) Positioning of items.
Marketers know that for every extra minute of 'quality' time that
one spends in a store, there is a high degree of likelihood that that
person will spend one extra dollar in that store (research states that
this is true for large retail stores). With this in mind, Walmart can
place Barbie dolls in one corner (Toys section) and Candy in a section
that is further away from the Toys section. And, in the path that lies
between these two sections, place special 'Kid' items - which may
be either promotion items, high margin items, Walmart's own brand
items, etc.
B) Co-packaging or Co-positioning
As you suggested, co-packaging is definitley a solution. But, this could
have the adverse effect of ultimately bringing down the sales of the dolls -
because the packaged items will cost more than the price of an individual
item. Co-packaging can be helpful in promotional cases, where the candy
may be being given away for free.
Co-positioning allows us to increase the sales of the follower items - in
this case candy. Barbie dolls and candy may be placed together or
very close to each other. The classic case of chips and dips is a good
example of this - where special racks were made in the chips section
to accomodate dips which ultimately helped increase the sales of dips.
C) The Bottom-Line issue.
Consider the following ...
If the profit margin on the leader item (A) is $1 and that of the follower (B)
is 75 cents. And, say, the confidence interval is known to be 60%.
Suppose that we sell 10,000 of A everyday. Thus, this amounts to a
sale of 6,000 of B everyday. Suppose we have a promotion offering
A on sale reducing our margin to 95 cents. At the same time we increase
our margin on B to 85 cents. By our sale, we increase our sales to 11,000
of A everyday (which amounts to 6,600 of B).
Pre-sale scenario ....
Profit on A = 10,000 * 1 = $10,000 per day
Profit on B = 6,000 * 0.75 = $4,500
Total Profit = $14,500
On-sale scenario ....
Profit on A = 11,000 * 0.95 = $10,450 per day
Profit on B = 6,600 * 0.85 = $5,610 per day
Total Profit = $16,060
Thus, by carefully adjusting promotions and prices, we can better our
bottom line.
D) The Advertising Lesson.
Retail stores advertise a lot - their advertisement budgets touch the sky.
Most of this money is going down the drain. These inflated advertisement
expenditures are begging to be better managed. The lesson to be learnt
here is that one should not advertise both a leader and a follower. By
advertising the leader, the follower automatically posts an increase in its'
sales. By using these rules for better advertisement management, companies
can either save on advertisement, or make their dollar reach out more than
before.
Previous2NextTop
Date: Fri, 26 Dec 1997 22:59:24 -0800
From: Ronny Kohavi ronnyk@starry.engr.sgi.com
Subject: MineSet mailing list and article in Data Management Strategies
We have setup a mailing list for people interested in Silicon
Graphics' MineSet announcements and discussions.
To subscribe to the mineset_list mailing list, send e-mail to external-majordomo@postofc.corp.sgi.com
with the BODY (subject is ignored) containing one line:
subscribe mineset_list _your_email_address_here_
In addition, we would like to tell you about a very nice article
evaluating MineSet 2.0 that appeared in Data Management Strategies.
I examined MineSet about a year ago and was impressed with its
capabilities. MineSet 2.0 includes some important new additions that
make this very capable data mining tool even more impressive.
MineSet has exceptional data visualization capabilities. But more
important for data analysts, MineSet's data mining and data
visualization capabilities are tightly integrated with each other...
...MineSet's three-dimensional landscape format, which uses a
'fly-through' navigational format, presents uncovered patterns and
trends to the user in a manner that is intuitive yet avoids
cluttering and overwhelming the analyst.
MineSet's capabilities will appeal to a variety of analysts, ranging
from database marketing and brand managers to retail buyers and
stockbrokers on Wall Street. In addition, scientists and engineers
will love its data modeling, visualization, and animation facilities.
--
Previous3NextTop
Date: Sun, 21 Dec 97 08:37:25 -0800
From: 'Shivakumar Vaithaynathan' SHIV@almaden.ibm.com
Subject: 2nd notice: AI Review: Special Issue on Data Mining on the Internet
Due to several requests, the deadline for the following special issues
has been extended till the 30th of January.
Artificial Intelligence Review:
Special Issue on Data Mining on the Internet
The advent of the World Wide Web has caused a dramatic increase in usage
of the Internet. The resulting growth in on-line information combined
with the almost chaotic nature of the web necessitates the development
of powerful yet computationally efficient algorithms to track and tame
this constantly evolving complex system.
While traditionally the data mining community has dealt with
structured databases, web mining poses problems not only due to the
lack of structure, but also due to the intrinsic distributed nature of
the data. Furthermore, mining on the Internet involves also dealing
with multi-media content consisting of not only natural language
documents but also images, audio and video streams. Several
interesting and potentially useful applications have already been
developed by academic researchers and industry practitioners to address
these challenges. It is important to learn from these initial endeavors,
if we are to develop new algorithms and interesting applications.
The purpose of this special issue is to provide a comprehensive
state-of-the-art overview of the technical challenges and successes
in mining of the Internet. Of particular interest are papers
describing both the development of novel algorithms and applications.
Topics of interest could include but are not limited to:
* Resource Discovery
* Collaborative Filtering
* Information Filtering
* Content Mining (text, images, video, etc.)
* Information Extraction
* User Profiling
* Applications, e.g., one-to-one marketing
In addition to the call for full-length papers, we request that any
researchers working in this area submit abstracts and/or pointers to
recently published applications for the purposes of compiling a
comprehensive survey of the current state-of-the-art.
<
**** Instructions for submitting papers ***
Papers should be no more than 30 printed pages (approximately 15,000
words) with a 12-point font and 18-point spacing, including figures
and tables. Papers must not have appeared in, nor be under
consideration by other journals. Include a separate page specifying
the paper's title and providing the address of the contact author for
correspondence (including postal, telephone number, fax number, and
e-mail address). Send FOUR copies of each submission to the guest
editor listed below. Papers in ascii or postscript form may be
submitted electronically. Instructions for on-line submission are
given below.
==================================
Information For on-line submission
==================================
Kluwer Academic Publishers allows on-line submission
of scientific articles via ftp and e-mail. We will make
this system more user-friendly by incorporating it into our
KAPIS WWW server and use Netscape as the user-interface.
This is currently being prepared and will be implemented by
the end of this year. Below, please find the procedure that
should be used until then.
- an author sends an e-mail message to 'submit@wkap.nl'
containing the
following line
REQUEST SUBMISSIONFORM AIRE
AIRE = Artificial Intelligence Review (the 4-letter code that is used
at Kluwer)
- the author receives the electronic submission form (see attachment)
via e-mail with a dedicated file name filled in (and also the
information that is given at point 4: the journal's four-letter code plus
the full journal title)
- the author fills in the submission form and send it back to: 'submit@wkap.nl'
- at the same time, the author submits his/her article via anonymous ftp
at the following address: 'ftp.wkap.nl' in the subdirectory
INCOMING/SUBMIT, using the dedicated file name with an appropriate
extension
- at Kluwer, the article is registrated and taken into production in the
usual way
Papers Due: January 30, 1989
Acceptance Notification: April 1, 1998
Final Manuscript due: July 1, 1998
Guest Editor: Shivakumar Vaithyanathan, net.Mining, IBM Almaden Research Center,
650 Harry Road, San Jose, CA 95120
(408)927-2465 (Phone)
(408)927-2240 (Fax)
e-mail: shiv@almaden.ibm.com
Previous4NextTop
Date: Wed, 17 Dec 1997 21:32:45 +0800
From: Stephen Koo skoo@hkstar.com
Subject: Datamining articles in Chinese
[The following site has data mining related articles,
but in Chinese ! GPS]
Dear Gregory,
I am one of your Nuggets subscriber, and one of freelance journalist
writing datamining articles in Chinese. The articles are published in
local Chinese newspaper weekly. The audiences are general public and
computing professional and practioners. I know your site is almost an
official site for datamining sources. Please go to my site http://www.hkstar.com/~skoo,
if you find it useful and beneficial to
datamining community and technology sharing, please add my site into
your link. If you have any queries, please feel free to contact me.
For your information, ISoft has announced ALICE on the Web, which gives
complete Internet access to ALICE d'ISoft.
New Features:
Mine your data on a remote site
Export mining results
Visualize Decision Tree Model with editing functions on a remote workstation.
Other news: ISoft and Valoris Technologies have signed a
partnership agreement.
Valoris now possesses datamining competence and has a service and training
offer based on ISoft datamining solutions. In particular, Alice d'ISoft is
integrated to Valoris Technology Workshops.
Raphaelle Thomas
International Development Manager
Tel : + 33 (0)1 69 35 37 37
Fax : + 33 (0)1 69 35 37 39
Email : rthomas@isoft.fr
SCI'98-ISAS'98 FOCUS SYMPOSIUM ON KDD, Orlando, Florida, July 12-16, 1998
In the context of the 1998 WORLD MULTICONFERENCE ON SYSTEMICS, CYBERNETICS
AND INFORMATICS and the 1998 INTERNATIONAL CONFERENCE ON INFORMATION SYSTEMS
ANALYSIS AND SYNTHESIS, to be held in Orlando (Florida) next July 12-16, we
are organizing a FOCUS SYMPOSIUM on Knowledge Discovery in Databases (KDD).
The topics of interest include, but are not limited to:
* High dimensional data sets and data preprocessing
* Data and knowledge acquisition and representation
* Use of prior domain knowledge and re-use of discovered knowledge
* Algorithmic complexity, efficiency and scalability issues in data
mining
* Distributed discovering algorithms and parallel processing
* Unsupervised discovery
* Clustering techniques
* Probabilistic and statistical models and methods
* Uncertainty management
* Data mining tools
* Methods for evaluating subjective interestingness and utility
* Data and knowledge visualization
* Applications of KDD systems
Participants who wish to present a paper, are requested to submit a
condensed first draft including title, author name(s), affiliation(s),
e-mail address(es), together with an abstract (500-1500 words) by February
15th, 1998. Submissions must be sent by e-mail to any of the addresses of
the Focus Symposium Organizers.
All submitted abstracts will be reviewed on the basis of technical quality,
novelty, significance and clarity. Acceptance notifications will be done by
April 15th, 1998.
Final versions should be sent by May 15th, 1998. Accepted papers will be
included in proceedings.
AAAI/ICML-98 Workshop on
Learning for Text Categorization
to be held July 27, 1998 in Madison, WI
The enormous growth of on-line information, has led to a comparable
growth in the need for methods that help users organize such
information. One area in particular that has seen much recent
research activity is the use of automated learning techniques to
categorize text documents. Such methods are useful for addressing
problems including, but not limited to: keyword tagging, word sense
disambiguation, information filtering and routing, sentence parsing,
clustering of related documents and classification of documents into
pre-defined topics.
The aim of this workshop is to examine recent theoretical,
methodological, and practical innovations from the various communities
interested in text categorization. The workshop will cover recent
advances from such fields as Machine Learning, Bayesian Networks,
Information Retrieval, Natural Language Processing, Case-Based
Reasoning, Language Modeling and Speech Recognition. By analyzing the
different underlying assumptions and state-of-the-art methodologies
used in text categorization research, as well as successful
applications of this work, we hope to foster new interactions between
researchers in this area.
OBJECTIVES: This workshop will focus on refining the state-of-the-art
in applying machine learning (ML) techniques rather than documenting
application experiences. Our objective is to analyze existing
experience to extract guidelines for developing ML applications.
TOPICS: (including but not limited to)
* Frameworks for creating ML applications and reusing parts of
previously developed applications
* Methodologies for applying ML techniques
* The roles of knowledge necessary for applying ML
* Matching problem definitions to specific technique configurations
* Relating and characterizing ML techniques with problem types
* Embedding the ML application process in knowledge
acquisition and system development methodologies
* Comparing ML applications with applications of related techniques
* Approaches that combine human and automated learning agents
The ELITE Foundation (European Laboratory for Intelligent Techniques
Engineering) is pleased to announce EUFIT `98 - The 6th European
Congress on Intelligent Techniques and Soft Computing. EUFIT `98 aims
to bring together scientists and practitioners from academic,
governmental, and industrial institutions to discuss new developments
and results in the field of intelligent technologies.
The congress will take place in Aachen (Aix-la-Chapelle), Germany, on
September 7-10, 1998.
<>
Structure of the Congress
Tutorials: September 7, 1998
Conference: September 8 - 10, 1998
Exhibition: September 7 - 9, 1998
Working Groups: September 7 - 11, 1998
You are invited to show your interest for EUFIT `98 only by completing the
preregistration form, and returning it immediately.