Readable and Accurate Rulesets with ORGA

Authors:
Md Nor Daud;David Corne
Affiliations:
School of Mathematics and Computer Sciences, Heriot-Watt University, Edinburgh, UK EH14 8AS;School of Mathematics and Computer Sciences, Heriot-Watt University, Edinburgh, UK EH14 8AS
Venue:
Proceedings of the 10th international conference on Parallel Problem Solving from Nature: PPSN X
Year:
2008

Citing 11
Cited 0

Boolean Feature Discovery in Empirical Learning

Machine Learning
C4.5: programs for machine learning

C4.5: programs for machine learning
Very Simple Classification Rules Perform Well on Most Commonly Used Datasets

Machine Learning
Induction of ripple-down rules applied to modeling large databases

Journal of Intelligent Information Systems
Data Mining and Knowledge Discovery with Evolutionary Algorithms

Data Mining and Knowledge Discovery with Evolutionary Algorithms
The Power of Decision Tables

ECML '95 Proceedings of the 8th European Conference on Machine Learning
Generating Accurate Rule Sets Without Global Optimization

ICML '98 Proceedings of the Fifteenth International Conference on Machine Learning
A hybrid decision tree/genetic algorithm method for data mining

Information Sciences: an International Journal - Special issue: Soft computing data mining
Data Mining: Practical Machine Learning Tools and Techniques, Second Edition (Morgan Kaufmann Series in Data Management Systems)

Data Mining: Practical Machine Learning Tools and Techniques, Second Edition (Morgan Kaufmann Series in Data Management Systems)
Cost-sensitive classification: empirical evaluation of a hybrid genetic decision tree induction algorithm

Journal of Artificial Intelligence Research
Hybrid learning using genetic algorithms and decision trees for pattern classification

IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 1

Quantified Score

Hi-index	0.00

Visualization

Abstract

A key task for data mining is to produce accurate and descriptive models. `Human readable' models are often necessary to enable understanding, potentially leading to further insight, and also inducing trust in the user. Rules, or decision trees (if not too numerous or large) are readable, unlike, for example SVM models. However, descriptiveness and accuracy normally conflict; a challenge is to find algorithms that have both high accuracy and high readability. We introduce ORGA (Optimized Ripper using Genetic Algorithm) which hybridizes evolutionary search with the RIPPER ruleset algorithm. RIPPER is effective at producing accurate and readable rulesets, and we show that ORGA provides significant further improvement. ORGA outperforms overall a suitable set of comparative algorithms including implementations of RIPPER, C4.5 and PART. On a majority of the datasets, ORGA's outperformance of the other algorithms is spectacular, and it is rarely dominated in terms of both accuracy and readability.