DM: clickstream Sequence Analysis

CaprI (Clementine a priori Intervals) is a data mining algorithm, used
within the Clementine environment, that discovers different types of
sequences across records (and hence time).  

Typical sequential patterns that can be found in datasets using CaprI are:
· Products bought by customers across multiple transactions;
· Financial transactions made by a business in a fiscal year;
· Regularly occurring browsing behaviour of visitors to an electronic
commerce web site; or
· Frequent sequences of the chemical bases that make up human DNA.

A 'sequence' field must also be specified in order for CaprI to know which
records go together to form a sequence.  Typical Sequence fields are
attributes such as user_id, account_number and so on.  The number of
distinct values for the Sequence field provides the total number of 'sets'
(e.g. the number of individual customers in a retail transaction dataset)
from which the sequences are derived.
The user will also have the option of specifying another of the INPUT fields
as a 'Time field'; this will be used to sort the records into time order.  
INPUT fields can be both numeric and symbolic, but symbolic field types are
preferred.
CaprI also enables the user to incorporate knowledge of what they are
looking for in sequences, by allowing the user to specify features like the
start and end items for a sequence, sequence length and time constraints
such as maximum time allowed between items in the sequence. It is also
possible to specify start and/or sequence end points. So that users can
explore all paths to say the buy page or all paths from a particular start
page.


Regards


Trevor Lilley 
SPSS

-----Original Message-----
From: Scott Nicholson [mailto:scottnicholson@yahoo.com] 
Sent: Monday, February 07, 2000 7:52 AM
To: datamine-l@nautilus-sys.com
Subject: DM: Software suggestion for clickstream data mining?


Does anyone have a suggestion for good software to aid
in clickstream analysis?  I'm looking for something
that goes beyond the typical analysis and provides
pathways of clicks, allowing the grouping of customers
by what they tend to look at.  

Thanks,
Scott Nicholson
__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

• Prev by Date: Re:DM: CHAID
• Next by Date: Re: DM: CHAID
• Prev by thread: Re: DM: Datamining in Internet...
• Next by thread: DM: PKDD-2000 Workshops and Symposia Program - CALL FOR PROPOSALS
• Index(es):
  • Date
  • Thread