Nová GUHA-procedura ETree-Miner v systému LISp-Miner

This paper presents a new GUHA-procedure called ETree-Miner which utilizes a new concept of so called exploration trees and which was recently implemented into LISp-Miner, an academic system for KDD developed at Faculty of Informatics, University of Economics Prague. Main differences between exploration trees and classical decision trees are explained. A brief summary of GUHA-method and LISp-Miner is included. Selected details of implementation ETree-Miner are described and finally, relations to other projects are mentioned. A de-facto standard PMML for description of data-mining models is used for communications with external systems.