摘要
We consider quantile functional regression with a functional part and a scalar linear part. We establish the optimal prediction rate for the model under mild assumptions in the reproducing kernel Hilbert space (RKHS) framework. Under stronger assumptions related to the capacity of the RKHS, the non-functional linear part is shown to have asymptotic normality. The estimators are illustrated in simulation studies.
学科与关键词
关键词
Convergence rateprediction riskquantile regressionrademacher complexity
- Sparse high-dimensional semi-nonparametric quantile regression in a reproducing kernel Hilbert spaceWang, Yue; Zhou, Yan; Li, Rui*; Lian, HengCOMPUTATIONAL STATISTICS & DATA ANALYSIS, 2022, 168: 107388.被引 5
- Kernel-based estimation for partially functional linear model: Minimax rates and randomized sketchesLv, Shaogao*; He, Xin; Wang, JunhuiJOURNAL OF MACHINE LEARNING RESEARCH, 2023, 24.
- Functional additive expectile regression in the reproducing kernel Hilbert spaceLiu, Yuzi; Peng, Ling; Liu, Qing; Lian, Heng; Liu, Xiaohui*Journal of Multivariate Analysis, 2023, 198: 105214.
