Tree-based classifier ensembles for early detection method of diabetes: an exploratory study

摘要: Diabetes is a lifestyle-driven disease which has become critical health issue worldwide. In this paper, we conduct an exploratory study about early detection method of diabetes mellitus using various ensemble learning techniques. Eight tree-based machine algorithms, i.e. classification and regression tree, decision tree (C4.5), reduced error pruning random naive Bayes functional best-first logistic model are employed as base classifier in five different ensembles, bagging, boosting, subspace, DECORATE, rotation forest. The performance ensembles classifiers thoroughly benchmarked on three real-world datasets term area under receiver operating characteristic curve metric. Finally, assess the differences among several statistical significant tests. We contribute to existing literature regarding extensive benchmark for disease.