Statistics and Its Interface

Volume 3 (2010)

Number 4

Robust neural network with applications to credit portfolio data analysis

Pages: 437 – 444

DOI: https://dx.doi.org/10.4310/SII.2010.v3.n4.a2

Authors

Yijia Feng (Department of Statistics, The Pennsylvania State University, University Park, Penn., U.S.A.)

Runze Li (Department of Statistics, The Pennsylvania State University, University Park, Penn., U.S.A.)

Agus Sudjianto (Bank of America, Charlotte, North Carolina, U.S.A.)

Yiyun Zhang (Novartis Oncology, OGD BDM Gbl Bios Full Dev, Florham Park, New Jersey, U.S.A.)

Abstract

In this article, we study nonparametric conditional quantile estimation via neural network structure. We proposed an estimation method that combines quantile regression and neural network (robust neural network, RNN). It provides good smoothing performance in the presence of outliers and can be used to construct prediction bands. A Majorization- Minimization (MM) algorithm was developed for optimization. Monte Carlo simulation study is conducted to assess the performance of RNN. Comparison with other nonparametric regression methods (e.g., local linear regression and regression splines) in real data application demonstrate the advantage of the newly proposed procedure.

Keywords

conditional quantile, nonparametric regression

2010 Mathematics Subject Classification

62G35, 62P99

Published 1 January 2010