654
Views
13
CrossRef citations to date
0
Altmetric
Articles

Geographically Weighted Elastic Net: A Variable-Selection and Modeling Method under the Spatially Nonstationary Condition

& ORCID Icon
Pages 1582-1600 | Received 01 Aug 2016, Accepted 01 Oct 2017, Published online: 19 Mar 2018
 

Abstract

This study develops a linear regression model to select local, low-collinear explanatory variables. This model combines two well-known models: geographically weighted regression (GWR) and elastic net (EN). The GWR model posits that the regression coefficients vary as a function of location and focuses on solving the problem of explaining the relationships under the spatially nonstationary condition, which a global model cannot solve. GWR cannot fulfill the task of variable selection, however, which is problematic when there are many explanatory variables with nonnegligible multicollinearity. On the other hand, the EN model is a member of the regulated regression family. EN can trim the number of explanatory variables and select the most important ones by adding penalty terms in its cost function, and it has been proven to be robust under the high-multicollinearity condition. The EN model is a global model, however, and does not consider the spatial nonstationarity. To overcome these deficiencies, we proposed the geographically weighted elastic net (GWEN) model. GWEN uses the kernel weights derived from GWR and applies EN locally to select variables for each geographical location. The result is a set of locally selected, low-collinear explanatory variables with spatially varying coefficients. We demonstrated the GWEN method on a data set relating population changes to a set of social, economic, and environmental variables in the Lower Mississippi River Basin. The results show that GWEN has the advantages of both the high prediction accuracy of GWR and the low multicollinearity among explanatory variables of EN.

本研究发展一个线性迴归模型来选择在地且低共线性的解释变因。此一模型结合了两个知名的模型: 地理加权迴归 (GWR) 与弹性网络 (EN)。GWR 模型假定迴归系数作为区位函数而有所变化, 并聚焦解释空间非静态条件下的关系之问题, 而该问题无法由全球模型解决。但 GWR 无法完成变因选择的任务, 因此当具有诸多无法忽略的多重共线性之解释变因时便会产生问题。此外, EN 模型是调节迴归家族中的一员。EN 能够通过在其成本函数中增加处罚条款, 缩减解释变因的数量, 并选择最重要的变因, 且已被证实在高度多重共线性的条件下是有效的。但 EN 模型是全球模型, 而且不考量空间非静止性。为了克服这些缺陷, 我们提出地理加权弹性网络 (GWEN) 模型。 GWEN 运用 GWR 衍生的核加权, 并将 EN 运用至地方, 以选择各地理区位的变因。该结果是一组地方选择的低度共线解释变因, 并具有空间变异的系数。我们在一组将人口变迁连结至密西西比河流域下游的社会、经济与环境变因组的数据集来展现 GWEN 模型。该结果显示, GWEN 同时具有 GWR 高度预测准确性以及 EN 解释变因的低度多重共线性之优势。

Este estudio desarrolla un modelo de regresión lineal para seleccionar variables explicativas de colineal bajo. Este modelo combina dos modelos bien conocidos: la regresión geográficamente ponderada (GWR) y la red elástica (EN). El modelo GWR plantea que los coeficientes de regresión varían como una función de localización y se enfoca en resolver el problema de explicar las relaciones bajo la condición espacialmente no estacionaria, que un modelo global no puede resolver. Sin embargo, la GWR no puede cumplir la tarea de selección de variables, que resulta problemática cuando hay muchas variables explicativas con multicolinearidad no desdeñable. Por otra parte, el modelo EN hace parte de la familia de regresión regulada. El EN puede recortar el número de variables explicativas y seleccionar las más importantes añadiendo términos de sanción en su función de costo, además de haber resultado robusto bajo la condición de alta multicolinearidad. Sin embargo, el EN es un modelo global que no considera la no estacionalidad espacial. Para remediar estas deficiencias, proponemos el modelo de la red elástica geográficamente ponderada (GWEN). El GWEN usa los pesos kernel derivados de la GWR y aplica localmente el EN para la tarea de seleccionar variables en cada localización geográfica. El resultado es un conjunto de variables explicativas de colineal bajo localmente seleccionadas con coeficientes que varían espacialmente. Hicimos una demostración del método GWEN sobre un conjunto de datos que relacionan los cambios de población con un conjunto de variables sociales, económicas y ambientales en la Cuenca del Bajo Río Misisipi. Los resultados muestran que el GWEN reúne las ventajas tanto de la alta exactitud de predicción de la GWR como la baja multicolinearidad propia de las variables explicativas del EN.

Additional information

Funding

This article is based on work supported by two research grants from the U.S. National Science Foundation: one under the Dynamics of Coupled Natural Human Systems (CNH) Program (Award No. 121211) and the other under the Coastal Science, Engineering and Education for Sustainability (Coastal SEES) Program (Award No. 1427389). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funding agencies.

Notes on contributors

Kenan Li

KENAN LI is a Postdoctoral Research Associate in the Department of Preventive Medicine (Bio-stats Division) at University of Southern California, Los Angeles, CA 90033. E-mail: [email protected]. His research interests include agent-based modeling of the sociosystems, spatial and temporal analysis of the environmental health issues, and developing integrated sensor monitoring systems for epidemiological studies.

Nina S. N. Lam

NINA S. N. LAM is Professor and E. L. Abraham Distinguished Professor of Louisiana Environmental Studies in the Department of Environmental Sciences at Louisiana State University, Baton Rouge, LA 70803. E-mail: [email protected]. Her research interests include GIScience, remote sensing, environmental health, resilience, and sustainability.

Log in via your institution

Log in to Taylor & Francis Online

PDF download + Online access

  • 48 hours access to article PDF & online version
  • Article PDF can be downloaded
  • Article PDF can be printed
USD 53.00 Add to cart

Issue Purchase

  • 30 days online access to complete issue
  • Article PDFs can be downloaded
  • Article PDFs can be printed
USD 312.00 Add to cart

* Local tax will be added as applicable

Related Research

People also read lists articles that other readers of this article have read.

Recommended articles lists articles that we recommend and is powered by our AI driven recommendation engine.

Cited by lists all citing articles based on Crossref citations.
Articles with the Crossref icon will open in a new tab.