Variable selection for general transformation models with right censored data via nonconcave penalties

In this paper, we consider variable selection for general transformation models with right censored data via nonconcave penalties. We will conduct the variable selection by maximizing the penalized log-marginal likelihood function. In the proposed variable selection procedures, we not only can select significant variables and but also are able to estimate corresponding effects simultaneously. With proper penalties and some conditions, we show that the resulting penalized estimates are consistent and enjoy oracle properties. We will illustrate our proposed variable selection procedures through some simulation studies and a real data application.