The K Nearest Neighbors Estimation of the Conditional Hazard Function for Functional Data

Abstract

In this paper, we study the nonparametric estimator of the conditional hazard function using the k nearest neighbors (k-NN) estimation method for a scalar response variable given a random variable taking values in a semi-metric space. We give the almost complete convergence (its corresponding rate) of this estimator and we establish the asymptotic normality. Then the effectiveness of this method is exhibited by a comparison with the kernel method estimation given in Ferraty et al. ([12]) and Laksaci and Mechab ([15]) in both cases simulated data and real data.

Keywords

• functional data
• nonparametric regression
• k-NN estimator
• the conditional hazard function
• rate of convergence
• random bandwidth