A General Data Mining Methodology Based on a Weighted Hierarchical Adaptive Voting Ensemble (WHAVE) Machine Learning Method
Clemen Deng and Marek Perkowski
This paper presents a general data mining methodology based on a novel Weighted Hierarchical Adaptive Voting Ensemble (WHAVE) machine learning (ML) method. It was constructed using three individual ML methods based on Multiple-Valued Logic: Disjunctive Normal Form (DNF) rule based method, Decision Trees, Naïve Bayes, and one method based on continuous representation: Support Vector Machines (SVM). The WHAVE method was demonstrated in applications for breast cancer, heart disease detection and stock market prediction with accuracies of 99.8%, 96.7% and 95.2% respectively. Results were compared with other methods and show that the WHAVE method accuracy is noticeably higher than those of the individual ML methods tested. This paper demonstrates the advantage of this new machine learning methodology based on a hierarchical ensemble.
Keywords: Data mining; machine learning; ensemble; majority voting system; multi-valued logic; information gain.