k-Nearest Neighbour method in functional nonparametric regression
k-Nearest Neighbour method in functional nonparametric regression
复制标题
DOI：
10.1080/10485250802668909
复制
发表时间：
2009-01-01
影响因子：
1.2
通讯作者：
Vieu, Philippe
中科院分区：
文献类型：
--
作者：
Burba, Florent;Ferraty, Frederic;Vieu, Philippe
本文的目的是研究非参数函数回归中的 k 最近邻 (kNN) 方法。我们提出了 kNN 核估计器的渐近特性：几乎完全收敛及其速率。然后，我们首先在模拟数据集上与传统核方法进行比较，然后在真实的化学计量示例上进行比较，以说明该方法的有效性。我们还在本文中介绍了一种重要的技术工具，该工具在除我们的情况之外的许多其他情况下都可能有用。
The aim of this article is to study the k-nearest neighbour (kNN) method in nonparametric functional regression. We present asymptotic properties of the kNN kernel estimator: the almost-complete convergence and its rate. Then, we illustrate the effectiveness of this method by comparing it with the traditional kernel approach first on simulated datasets and then on a real chemometrical example. We also present in this article an important technical tool which could be useful in many other situations than ours.