Statistics and Its Interface
Volume 7 (2014)
Nonparametric quantile regression models via majorization minimization-algorithm
Pages: 235 – 240
In this paper, we apply the Majorization Minimization (MM)-algorithm to deal with the computational problem of the smoothing nonparametric quantile regression. We show that the proposed MM-algorithm possesses the descent property, and the estimator obtained by the proposed algorithm is smooth. Simulation studies demonstrate that the estimator based on our proposed method is more robust and efficient than the estimator based on the mean smoothing regression and the estimator proposed by Nychka et al. (1995) with GCV scores. Finally, we apply the proposed methodology to analyze the dataset about bone density (BMD) in adolescents.
MM-algorithm, nonparametric quantile regression, robustness
2010 Mathematics Subject Classification
Primary 62F35. Secondary 62G08.