A DEA approach for model combination

This paper proposes a novel Data Envelopment Analysis (DEA) based approach for model combination. We first prove that for the 2-class classification problems DEA models identify the same convex hull as the popular ROC analysis used for model combination. For general k-class classifiers, we then develop a DEA-based method to combine multiple classifiers. Experiments show that the method outperforms other benchmark methods and suggest that DEA can be a promising tool for model combination. Categories and Subject Descriptors H.2.8 [Database Management]: Applications ? Data Mining; I.2.6 [Artificial Intelligence]: Learning General Terms Algorithms, Experimentation Keywords Model Combination, Data Envelopment Analysis, ROC

• Type: Conference
• Year: 2004
• Where: KDD
• Authors: Zhiqiang Zheng, Balaji Padmanabhan, Haoqiang Zheng
• Keywords: Data Envelopment Analysis, Data Mining, KDD 2004, Keywords Model Combination, Model Combination