“ Missing is Useful ” : Missing Values in Cost - sensitive Decision Trees 1

Many real-world datasets for machine learning and data mining contain missing values, and much previous research regards it as a problem, and attempts to impute missing values before training and testing. In this paper, we study this issue in cost-sensitive learning that considers both test costs and misclassification costs. If some attributes (tests) are too expensive in obtaining their values, it would be more cost-effective to miss out their values, similar to skipping expensive and risky tests (missing values) in patient diagnosis (classification). That is, “missing is useful” as missing values actually reduces the total cost of tests and misclassifications, and therefore, it is not meaningful to impute their values. We discuss and compare several strategies that utilize only known values and that “missing is useful” for cost reduction in cost-sensitive decision tree learning. 1 This work is partially supported by Australian large ARC grants (DP0343109 and DP0559536), a China NSFC major research Program (60496321), and a China NSFC grant (60463003). • Shichao Zhang is with the Department of Computer Science at Guangxi Normal University, Guilin, China; and with the Faculty of Information Technology at University of Technology Sydney, PO Box 123, Broadway, Sydney, NSW 2007, Australia; zhangsc@ it.uts.edu.au. • Zhenxing Qin is with the Faculty of Information Technology at University of Technology Sydney, PO Box 123, Broadway, Sydney, NSW 2007, Australia; zqin@ it.uts.edu.au. • Charles X. Ling, Shengli Sheng are with the Department of Computer Science at The University of Western Ontario, London, Ontario N6A 5B7, Canada; {cling, ssheng}@ csd.uwo.ca.