zakimjz / schism

SCHISM finds interesting subspace clusters

SCHISM software

Authors - Karlton Sequeira, Mohammed Zaki (2004)

SCHISM finds interesting subspace clusters.

K. Sequeira, M. J. Zaki. SCHISM: A New Approach for Interesting Subspace Mining. In the Proceedings of the Fourth IEEE Conference On Data Mining. 2004.

Karlton Sequeira and Mohammed J. Zaki. SCHISM: a new approach to interesting subspace mining. International Journal of Business Intelligence and Data Mining, 1(2):137–160, 2005. doi:10.1504/IJBIDM.2005.008360.

FUNCTIONALITY

dataGen creates horizontal format ASCII high-dimensional datasets having embedded subspaces

convertData converts horizontal format ASCII high-dimensional datasets output by dataGen to a suitable format i.e. vertical/binary/IBM/WEKA/LDR/MAFIA having embedded subspaces (Only IBM formats have been recently tested)

schism finds the embedded subspaces

RUNNING THEM

Command line arguments for convertData and dataGen can be found by running them without any arguments.

typically run them as

    dataGen -o /tmp/schism/data/swhatever.ha -d 60

This creates a horizontal ASCII dataset with default parameters and 60 dims.

    convertData -i /tmp/schism/data/swhatever -d 60

This creates a IBM format dataset with default parameters from the horizontal ASCII file earlier created.

    schism -i /tmp/schism/data/swhatever.ibm

This mines the IBM format file for embedded subspaces using default parameters

It prints "<Execution time(minutes)> <# interesting subspaces found> ". To print the actual subspaces, add parameter " -o 1".

In order to run these programs using as few parameters involves suffixing the horizontal ASCII file by ".ha" and placing all new files in the /tmp/schism/data directory

Example use may be seen in the shell scripts dataSCHISM.sh and testSCHISM.sh. Paths must be appropriately changed to run the scripts.