Cost sensitive SVM with non-informative examples elimination for imbalanced postoperative risk management problem

Abstract

In this paper we propose a novel combined approach to solve the imbalanced data issue in the application to the problem of the post-operative life expectancy prediction for the lung cancer patients. This solution makes use of undersampling techniques together with cost-sensitive SVM (Support Vector Machines). First, we eliminate non-informative examples by applying Tomek links together with onesided selection. Second, we take advantage of using cost-sensitive SVM with penalty costs calculated respecting cardinalities of minority and majority examples. We evaluate the presented solution by comparing the performance of our method with SVM-based approaches that deal with uneven data. The experimental evaluation was performed on reallife data from the postoperative risk management domain.