T3: A Classification Algorithm for Data Mining
This paper describes and evaluates T3, an algorithm that builds trees of depth at most three, and results in high accuracy whilst keeping the size of the tree reasonably small. T3 is an improvement over T2 in that it builds larger trees and adopts a less greedy approach. T3 gave better results than both T2 and C4.5 when run against publicly available data sets: T3 decreased classification error on average by 47% and generalisation error by 29%, compared to T2; and T3 resulted in 46% smaller trees and 32% less classification error compared to C4.5. Due to its way of handling unknown values, T3 outperforms C4.5 in generalisation by 99% to 66%, on a specific medical dataset.
KeywordsClassification Accuracy Classification Algorithm Tree Size Classification Error Continuous Attribute
Unable to display preview. Download preview PDF.
- 1.Aha, D.W., Breslow, L.A: Comparing Simplification Procedures for Decision Trees on an Economics Classification, NRL/FR/5510 98-9881, (Technical Report AIC-98-009), May 11, 1998.Google Scholar
- 2.Auer, P. Holte, R.C., Maass, W.: Theory and Applications of Agnostic PAC-Learning with Small Decision Trees, Proc. 12th Int’l Machine Learning Conf. San Francisco, Morgan Kaufmann 1995, pp. 21–29.Google Scholar
- 3.Breslow, L., Aha, D.W.: Comparing Tree-Simplification Procedures, Proc. 6th Int’l Workshop Artificial Intelligence and Statistics, Ft. Lauderdale, 1997, pp. 67–74.Google Scholar
- 4.Ganti, V., Gehrke, J., Ramakrishnan, R.: Mining Very Large Databases, IEEE Computer, Special issue on Data Mining, August 1999.Google Scholar
- 5.Kohavi, R., Sommerfield, D., Dougherty, J.: Data Mining using MLC++: A Machine Learning Library in C++, Tools with AI, 1996.Google Scholar
- 6.Murthy, S., Saltzberg, S.: Decision Tree Induction: How effective is the Greedy Heuristic?, Proc. 1st Int’l Conf. on KDD and DM, 1995, pp. 156–161.Google Scholar
- 7.Quinlan, J.R.: C4.5: Programs for Machine Learning, San Mateo, Morgan Kaufmann, 1993.Google Scholar