T3: A Classification Algorithm for Data Mining

• Authors: Christos Tjortjis;John Keane
• Venue: IDEAL '02 Proceedings of the Third International Conference on Intelligent Data Engineering and Automated Learning
• Year: 2002

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.