Keywords: Additive rate model, Event history study, Recurrent event data, Variable selection
This paper discusses regression analysis of recurrent event data with the focus on simultaneous parameter estimation and variable selection and clustering. Recurrent event data often occur in many areas such as medical studies and social sciences and a great deal of literature has been established for their analysis (Andersen et al., 1993; Cook and Lawless, 2007; Lin et al., 2000; Schaubel et al., 2006). However, only limited research exists for the variable selection in the context of recurrent event data. For the problem, we present a broken adaptive ridge regression approach that can not only allow one to perform parameter estimation and variable selection simultaneously but also have the clustering effect when covariates are highly correlated. The asymptotic properties of the proposed approach including the oracle property are established, and an extensive simulation study is performed and indicates that the method works well for practical situations. An application is also provided.