Subgroup discovery toolkit for Orange
Description
The subgroup discovery toolkit for Orange implements three algorithms for subgroup discovery: SD, CN2-SD and Apriori-SD, two visualization methods: the BAR and the ROC visualization and six evaluation measures for subgroup discovery.
It is distributed free under GPL and can be downloaded from this web page.
Requirements
One needs to have Orange installed and working. Orange is
available for download at here.
In order for the PMML functionality to work, one also needs to have
pyxml installed. The version for Python 2.5 is available
here.
Download
Zip file: SubgroupDiscovery_v1.0.1.zip (released on November 3, 2008)
Installation
- Unzip the downloaded file into the directory OrangeWidgets, which is usually placed here: C:\Python25\Lib\site-packages\orange\OrangeWidgets\.
- Run OrangeCanvas.
- In Orange canvas, select from the menu "Options" the function "Rebuild Widget Registry". A new "tab" named SubgroupDiscovery should appear in Orange canvas and the tool is ready to be used.
Screenshots
Subgroup discovery schema 1:
Subgroup BAR vizualization from Schema 1:
Subgroup ROC vizualization from Schema 1:
Scatter plot vizualization from Schema 1:
Subgroup discovery evaluation schema:
Subgroup evaluation from the evaluation schema:
Authors
Petra Kralj Novak(1), Nada Lavrač(1)
Implemented by Petra Kralj Novak(1) with the help of the Orange team(2).
Vid Podpečan, Grega Podlesek, Robert Ravnik and Miha Rojko also contributed in the implementation.
- (1) Department of Knowledge Technologies,
Jožef Stefan Institute
Jamova 39, 1000 Ljubljana, Slovenia - (2) Artificial Intelligence Laboratory,
Faculty of Computer and Information Science,
University of Ljubljana
Tržaška 25, 1000 Ljubljana, Slovenia
Contact
Petra Kralj Novak
phone: +386 1 477 31 29
e-mail: Petra.Kralj.Novak@ijs.si