Thoughts on:
"Mining Sequential Patterns:
Generalizations and Performance Improvements",
Ramakrishnan Srikant, Rakesh Agrawal
IBM Almaden Research Center, 1996
I do not have an electronic copy of this paper.
An improvement over AprioriAll for mining sequential patterns is
presented. This could relate to our work. We could adapt the method
for numeric data, mine for patterns and then mine for association rules
using the patterns as values for the time series attributes. In fact,
this possibility is common among our background material.
The problem being solved in this work is similar to that presented in
the Inter-Transactional rule mining work by Tung, Lu, Han, and Feng.
|