Identification of model order and number of neighbors for k-nearest neighbor resampling
Identification of model order and number of neighbors for k-nearest neighbor resampling
复制标题
DOI：
10.1016/j.jhydrol.2011.04.024
复制
发表时间：
2011-07
影响因子：
6.4
通讯作者：
Taesam Lee;T. Ouarda
中科院分区：
文献类型：
--
作者：
Taesam Lee;T. Ouarda
在水文气象学中，k-最近邻回归（KNNR）是最常用的补充短历史记录的随机模型之一。在KNNR模型中，需要选择模型阶数（d）和最近邻数（k）。传统上，一直使用规范性选择（k=n1/2，其中是记录长度），并且没有提供实用的解决方案。另一种适用的方法是广义交叉验证（GCV）。然而，据报道，在文献中，GCV是不切实际的选择Dandin的KNNR模型。在目前的研究中，我们提出了一种基于赤池信息准则（AIC）的方法来选择dandk。所提出的方法进行了验证，一些模拟数据集，并应用到科罗拉多河系统的案例研究。结果表明，所提出的基于AIC的方法是一个强大的模型选择ofandk。在模拟研究中，该模型特别导致选择与模拟数据集的真实的阶相同的模型阶。在案例研究中也给出了可接受的k值。
Among the various stochastic models used in hydrology and meteorology, thek-nearest neighbor resampling (KNNR) has been one of the most common alternatives to supplement the short historical records. In the KNNR model one needs to select the model order (d) and the number of nearest neighbors (k). Traditionally, the prescriptive selection (k=n1/2wherenis the record length) has been used forkand no practical solutions were provided to choosed. Another applicable approach is generalized cross-validation (GCV). However, it has been reported in the literature that GCV is not practical for the selection ofdandkin the KNNR model. In the current study we propose an approach to selectdandkbased on the Akaike information criterion (AIC). The proposed approach was validated on a number of simulated datasets and applied to the case study of the Colorado River system. The results indicate that the proposed AIC-based approach represents a robust model for the selection ofdandk. In the simulation study, the model led particularly to the selection of the same model orders as the real orders of the simulated datasets. It also gave acceptablekvalues in the case study.