Abstract:
|
Support vector machines (SVM) have demonstrated utility for a wide variety of classification tasks. A key feature of SVMs is that they allow for construction of both linear and non-linear decision rules which can yield better prediction when the data are complex, as is frequently the case in biomedical studies. A practical challenge for SVMs, as with many other classification methods, lies in the accommodation of missing data which commonly occur in real data applications due to imperfect data collection. Currently, many researchers rely on complete-case or imputation solutions which may introduce bias and lead to reduced classification accuracy. Therefore, we propose an EM-motivated solution to the incomplete data problem for SVMs which maintains the convex objective function and which allows the researcher to use the same software as the complete case solution. Simulations show that the proposed method often yields classification rules with higher accuracy than existing methods. We apply the approach to analyze data from HCV-TARGET, a longitudinal study of Hepatitis C patients.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.