DM: RE: Market Basket Analysis: request for references, algorithms & sof

[Abraham Meidan]

Hi Dirk,

WizSoft has 2 data mining products that are based on a proprietary 
association rule algorithm. 

One product, WizRule, reveals all the if-then rules in a data set 
with no limit as to the number of conditions. The product is 
designated for issuing data auditing (it points at the cases that 
deviate from the discovered rules as cases to be audited), but since 
it reveals all the associations, you can use it for market basket 
analysis as well.

The second product, WizWhy, reveals all if-then rules, and issues 
predictions on the basis of the discovered rules. You can use this 
product if you want to analyze the baskets in order to predict under 
what conditions one purchases a certain product (and you have to 
issue the predictions for each such product).

Both products run on Windows 95 / Win 95 and can read ASCII, Access 
or any other ODBC compliant database.

Both products are not limited by the number of fields or the number 
of records.

You can read about the two products and download working demos 
(limited just by the number of records) at: www.wizsoft.com

Best regards,

Abraham.

-----Original Message-----
From:   Van den Poel, Dirk [SMTP:Dirk.VandenPoel@econ.kuleuven.ac.be]
Sent:   ב מרץ 23 1998 12:32
To:     'datamine-l@nautilus-sys.com'
Subject:        DM: Market Basket Analysis: request for references, 
algorithms & software

Dear readers,

I am looking for techniques to perform a market basket analysis. This
problem involves the identification of what products are purchased
together by consumers. In essence it is an association problem. Can
anyone point to a technical discussion of the algorithms (references)
used to perform such an analysis?
Currently, I only found this functionality in SAS Enterprise miner. 
But
since our access to this software is still limited, I would like to 
ask
whether anyone knows of other implementations (maybe even
free/shareware)?

Thank you very much for any related feedback,

Dirk Van den Poel
K.U.Leuven / Department of Applied Economics
Naamsestraat  69
B-3000  Leuven
Belgium
Email: dirk.vandenpoel@econ.kuleuven.ac.be