Distributed Censored Quantile Regression

This article discusses an extension of censored quantile regression to a distributed setting. With the growing availability massive datasets, it is oftentimes arduous task analyze all data with limited computational facilities efficiently. Our proposed method, which attempts overcome this challenge, comprised two key steps, namely: (i) estimation both Kaplan-Meier estimator and model coefficients in parallel computing environment; (ii) aggregation coefficient estimations from individual machines. We study upper limit order number machines for environment, which, if fulfilled, guarantees that converges at comparable rate oracle estimator. In addition, we also provide further modifications systems including communication-facilitated adaptation sense Chen, Liu, Zhang nonparametric counterpart along direction Kong Xia regression. Numerical experiments are conducted compare existing estimators. The promising results demonstrate computation efficiency methods. Finally, practical concerns, cross-validation procedure developed can better select hyperparameters methodologies. Supplementary materials available online.