Publications

Many people in data mining have heard the famous "diapers and beer" example
(a significant number of people who buy diapers also tend to buy beer at
the same time).  It's often cited as an interesting association rule
discovered (or at least discoverable) from market basket data.

I heard an explanation of the origin of this example, from Lounette Dyer
via Ronny Kohavi, and I thought KDnuggets readers would be interested.
Lounette (loon at cogit dot com) gave me permission to forward it:
==============================
   The "diapers and beer" example was created by Tom Blishok who ran a
   retail consulting group for NCR.  They first did market basket for
   Osco Drugs in Chicago.  This pre-dated commercial availability of
   associations algorithms (1992?).  Tom had a team of a half dozen or
   so SQL hackers (I know a couple of them) creating queries that
   defined affinities that they thought might be in the data.  They
   found "cosmetics and greeting cards" and a few others.

   In the process of Tom creating materials for NCR to make the
   argument for why you want to house transaction data he needed a
   snappy example, so he dreamed up the "diapers and beer" example.  To
   the best of my knowledge it was never supported in any data that
   they analyzed.  Tom's original example had "diapers and beer"
   appearing in C-Store data which, to the best of my knowledge, they
   never analyzed at the transaction level.

   I have some of Tom's early writings and presentations on market
   basket analysis.  Its interesting that everyone thinks that MBA was
   invented by the assoc. algorithms camp, when actually the SQL camp
   had already been trying to find business value in it.
==============================

It's still a nice example, of course.  As they say, never let truth get
in the way of a good story.

Tom Fawcett
HP Labs
Palo Alto,

Editor Note: A search of old KDnuggets Issues 
 found two references to diapers and beer --
www.kdnuggets.com/news/96/n08.html
and
www.kdnuggets.com/news/96/n16.html

GPS