readme

A C++ library which finds associations within sets of items, using a fast in-memory algorithm

This is a test version of the alg1 data miner.

http://alg1.sf.net

To install:
tar zxf alg1-0.1.tar.gz
cd alg1-0.1
make

The output is an executable named "alg1".  Usage:  alg1 filename ntrans cutoff

filename:  name of binary file in format produced by IBM's synthetic data
generator.

ntrans:  number of sets to load from the file

cutoff:  the minimum number of sets in which a subset has to occur to
qualify as "frequent".

example:  alg1 data 1000 5

The files "data" and "pat" are included for demonstration; pat contains
a text listing of frequent sets which should be found by alg1, depending
on the cutoff used, and the frequency listed in pat.  These files
were produced by IBM Almaden's data generator, but there are a few details
about them which are not documented by IBM.  

alg1 was developed on an Intel-based Mandrake 8.1 system

Note:  alg1 only prints out frequent subsets if they have more than
3 items, to save screen space.  Better output handling is the next phase.

Feedback appreciated - Kendall Willets ([e-mail])