Title: Conditional distribution function estimation and bandwidth selection under censoring
Authors: Dimitrios Bagkavos - University of Ioannina (Greece) [presenting]
Montserrat Guillen - University of Barcelona (Spain)
Jens Perch Nielsen - City, University of London (United Kingdom)
Abstract: A smooth and continuous nonparametric estimate of the conditional cumulative distribution function for any arbitrary number of covariates in the right censored data setting is proposed. The estimate is obtained as the combination of the multivariate local linear smoothing of the data across all covariate dimensions and the kernel smoothing of the Kaplan-Meier estimate in the response direction. Implementation of the estimate in practice is facilitated by the development of a plug-in type bandwidth selector. The rule yields different amounts of smoothing for each coordinate direction, thus optimizing the estimate's performance across the full spectrum of its support. The asymptotic properties of all methodological contributions are quantified analytically and discussed in detail. Finally, numerical evidence is provided on the finite sample performance of the proposed methodological advances, and a real data analysis illustrates their benefits in practice.