2017/12/15 by Nicole Barthel, Candida Geerdens, Barthel, Nicole +5
Mathematics · Economics, Econometrics and Finance · #Statistical Distribution Estimation and Applications #Statistical Methods and Inference #Financial Risk and Volatility Modeling
paper · pdf · doi:10.48550/arxiv.1712.05845
In many time-to-event studies, the event of interest is recurrent. Here, the\ndata for each sample unit corresponds to a series of gap times between the\nsubsequent events. Given a limited follow-up period, the last gap time might be\nright-censored. In contrast to classical analysis, gap times and censoring\ntimes cannot be assumed independent, i.e. the sequential nature of the data\ninduces dependent censoring. Also, the recurrences typically vary between\nsample units leading to unbalanced data. To model the association pattern\nbetween gap times, so far only parametric margins combined with the restrictive\nclass of Archimedean copulas have been considered. Here, taking the specific\ndata features into account, we extend existing work in several directions: we\nallow for nonparametric margins and consider the flexible class of D-vine\ncopulas. A global and sequential (one- and two-stage) likelihood approach are\nsuggested. We discuss the computational efficiency of each estimation strategy.\nExtensive simulations show good finite sample performance of the proposed\nmethodology. It is used to analyze the association in recurrent asthma attacks\nin children. The analysis reveals that a D-vine copula detects relevant\ninsights, on how dependence changes in strength and type over time.\n