Support Vector Regression for Right Censored Data

We develop a unified approach for support vector machines for classification and regression in which the outcomes are a function of the survival times subject to right censoring. We present a novel support-vector regression algorithm that is adjusted for censored data. We provide finite sample bounds on the generalization error of the algorithm. We prove risk consistency for a wide class of probability measures and study learning rates. We apply the general methodology to estimation of the (truncated) mean, median, quantiles, and for classification problems. We present a simulation study that demonstrates the performance of the proposed approach.