ATDIDT
A versatile Data Mining software

Louis Wehenkel
Université de Liège, Institut Montefiore

Overview

On-line demonstration

Documentation

Obtain software

Download databases

Stochastic methods

ATDIDT is a general purpose Data Mining software developed at the University of Liège.

It is an interactive software, containing facilities for loading data bases from files and extending attributes through explicit programming or application of automatic learning (decision and regression tree induction, linear and generalized linear regression, nearest neighbor and clustering techniques, multilayer perceptrons). It comprises also various graphical tools for visualization of various types of diagrams, and results from automatic learning (decision tree and multilayer perceptron graphs).

ATDIDT is written partly in Emacs Lisp, partly in CommonLisp, and partly in C. Development started in 1988 and the total number of Lisp code lines is abpout 40,000.It uses the GNU Emacs editor as a front-end and various other free softwares as auxiliary tools (ghostview/ghostscript; xfig/transfig; gunzip).

This page provides all the information presently available about ATDIDT. This includes a documentation in pdf format, some data bases freely distributed and the executable versions distributed for demo and teaching purposes at the University of Liège, together with some on-line demonstration of related interest.

See Automatic Learning Techniques in power systems (L. Wehenkel, Kluwer Academic, 1998) for a description of the algorithms and of applications in the context of electric power systems.

On-line demonstration:

DT induction applet (Java 2.01).
An interactive decision tree induction applet written by Pierre Geurts.
You need a recent version of Netscape or Explorer to run this correctly.

Documentation of ATDIDT:

Short documentation (.pdf version) (90272 bytes)

User guide (.pdf version) (draft version) (1216794 bytes)

For conditions to obtain the software please contact Louis.Wehenkel@ulg.ac.be

Download databases:

German credit allocation DB (german.db). 109971 bytes.

Small database containing animal descriptions (zoo.db) &206 bytes.

Survival on Titanic catastrophe (titanic.db) 57374 bytes.

Tennis playing example used by Quinlan to illustrate ID3 (tennis.db) 523 bytes.

Small Electric power systems stability database (omib.db) 46574 bytes.

Iris database (iris.db) 7192 bytes.

Back links

Back Louis Wehenkel' home page

Back to Stochastic Methods home page

Back to Institut Montefiore home page

Back to the ULg home page

Last update: 4 October 1999