A Wavelet-Based Extension of Generalized Linear Models to Remove the Effect of Spatial Autocorrelation: Wavelet-based Extension of Generalized Linear Models

Biogeographical studies are often based on a statistical analysis of data sampled in a spatial context. However, in many cases standard analyses such as regression models violate the assumption of independently and identically distributed errors. In this article, we show that the theory of wavelets provides a method to remove autocorrelation in generalized linear models (GLMs). Autocorrelation can be described by smooth wavelet coefficients at small scales. Therefore, data can be decomposed into uncorrelated and correlated parts. Using an appropriate linear transformation, we are able to extend GLMs to autocorrelated data. We illustrate our new method, called the wavelet-revised model (WRM), by applying it to multiple regression with response variables conforming to various distributions. Results are presented for simulated data and real biogeographical data (species counts of the plant genus Utricularia [bladderworts] in grid cells throughout Germany). The results of our WRM are compared with those of GLMs and models based on generalized estimating equations. We recommend WRMs, especially as a method that allows for spatial nonstationarity. The technique developed for lattice data is applicable without any prior knowledge of the real autocorrelation structure. Los estudios biogeograficos basados en el analisis estadistico de datos de una muestra espacial son frecuentes en la literatura especializada. En muchos casos, sin embargo, los analisis estandar mas usados tales como los modelos de regresion violan el supuesto que asume una distribucion de errores independiente e identica. .En este articulo los autores muestran que la teoria de wavelets proporciona un metodo para eliminar la autocorrelacion espacial en modelos lineales generalizados. La autocorrelacion puede ser descrita por los coeficientes wavelet suavizados hacia/en escalas finas o pequenas. Por lo tanto, los datos se pueden descomponer en componentes correlacionados y componentes no correlacionados. Los autores aplican una transformacion lineal con el fin de modificar los modelos lineales generalizados y permitir asi su aplicacion y uso con datos autocorrelacionados. Para ilustrar el metodo propuesto, denominado modelo wavelet reformulado (wavelet-revised method), los autores lo aplican a una regresion multiple con variables dependientes correspondientes a diferentes distribuciones estadisticas. Los resultados se presentan en dos formatos: a) datos de simulaciones y b) datos biogeograficos reales (numero de especies de una especie vegetal de genero Utricularia [bladderworts] en distribuidas en la cuadricula o grilla geografica de Alemania). Los resultados del modelo de wavelets reformulado son comparables con los de modelos lineales generalizados y con los de modelos basados en ecuaciones de estimacion generalizadas (generalized estimating equations). Los autores recomiendan los modelos revisados por wavelets, sobre todo por ser un metodo que permite abordar patrones y procesos no estacionarios. La tecnica desarrollada para el latice de datos es aplicable sin ningun conocimiento previo de la estructura real de autocorrelacion. 生物地理研究通常基于空间场景中采样数据的统计分析。然而在许多情况下,诸如回归模型一类的标准分析法违背了独立性与误差同分布假设。本研究表明小波理论提供了在广义线性模型中消除空间自相关的方法。空间自相关可通过小尺度上的平滑小波系数加以描述。因此,源数据可被分解为相关的和不相关的部分。采用恰当的线性转换,可扩展广义线性模型以处理自相关数据。我们描述了被称为小波改进模型的新方法,并将其应用于响应变量服从不同分布的多元回归分析中。同时给出了基于模拟数据和实际生物地理数据(德国地区植物狸藻属物种数量的栅格数据)的分析结果。小波改进模型与其他的广义线性模型和基于广义估计方程的模型结果也进行了对比。我们推荐小波改进模型,特别是作为空间非平稳数据分析的方法。该技术可发展为应用于在现实空间自相关结构中没有任何先验知识的格网数据分析。

[1]  Noel A Cressie,et al.  Statistics for Spatial Data. , 1992 .

[2]  R. G. Davies,et al.  Methods to account for spatial autocorrelation in the analysis of species distributional data : a review , 2007 .

[3]  T. Simons,et al.  Spatial autocorrelation and autoregressive models in ecology , 2002 .

[4]  Gabriele Lohmann,et al.  Wavelet statistics of functional MRI data and the general linear model , 2003, Journal of magnetic resonance imaging : JMRI.

[5]  R. Dennis Cook,et al.  Generalized Linear Models , 2008 .

[6]  Todd R. Ogden,et al.  Wavelet Methods for Time Series Analysis , 2002 .

[7]  Carsten F. Dormann,et al.  A wavelet-based method to remove spatial autocorrelation in the analysis of species distributional data , 2008 .

[8]  I. Johnstone,et al.  Wavelet Threshold Estimators for Data with Correlated Noise , 1997 .

[9]  Timothy H. Keitt,et al.  SCALE‐SPECIFIC INFERENCE USING WAVELETS , 2005 .

[10]  Annette J. Dobson,et al.  An introduction to generalized linear models , 1991 .

[11]  Mike Rees,et al.  5. Statistics for Spatial Data , 1993 .

[12]  Daniel A. Griffith,et al.  Simulating Two-dimensional Autocorrelated Surfaces , 2010 .

[13]  T. Keitt Scaling Biodiversity: On the quantification of local variation in biodiversity scaling using wavelets , 2007 .

[14]  Mark R. T. Dale,et al.  Spatial Pattern Analysis in Plant Ecology: Spatial Pattern Analysis in Plant Ecology , 1999 .

[15]  A. Getis,et al.  Comparative Spatial Filtering in Regression Analysis , 2002 .

[16]  P. Legendre Spatial Autocorrelation: Trouble or New Paradigm? , 1993 .

[17]  Pierre Dutilleul,et al.  Stepwise Regression in Mixed Quantitative Linear Models with Autocorrelated Errors , 2006 .

[18]  Patrick Flandrin,et al.  Wavelet analysis and synthesis of fractional Brownian motion , 1992, IEEE Trans. Inf. Theory.

[19]  Wei Wei,et al.  The use of the Mexican Hat and the Morlet wavelets for detection of ecological patterns , 2005, Plant Ecology.

[20]  T. Spies,et al.  Characterizing canopy gap structure in forests using wavelet analysis , 1992 .

[21]  P. McCullagh,et al.  Generalized Linear Models , 1992 .

[22]  Jessica Gurevitch,et al.  Ecography 25: 601 -- 615, 2002 , 2022 .

[23]  Hulin Wu,et al.  Markov chain Monte Carlo for autologistic regression models with application to the distribution of plant species , 1998 .

[24]  David S. Stoffer,et al.  Time series analysis and its applications , 2000 .

[25]  S. Lele,et al.  A Composite Likelihood Approach to Binary Spatial Data , 1998 .

[26]  François G. Meyer Wavelet based estimation of a semi parametric generalized linear model of fMRI time-series , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[27]  D. Griffith Spatial Autocorrelation and Spatial Filtering: Gaining Understanding Through Theory and Scientific Visualization , 2010 .

[28]  B. Silverman,et al.  The Stationary Wavelet Transform and some Statistical Applications , 1995 .

[29]  K Y Liang,et al.  Longitudinal data analysis for discrete and continuous outcomes. , 1986, Biometrics.

[30]  Ingolf Kühn,et al.  Analyzing spatial ecological data using linear regression and wavelet analysis , 2008 .

[31]  E. Bullmore,et al.  Wavelet-Generalized Least Squares: A New BLU Estimator of Linear Regression Models with 1/f Errors , 2002, NeuroImage.

[32]  François G. Meyer Wavelet-based estimation of a semiparametric generalized linear model of fMRI time-series , 2003, IEEE Transactions on Medical Imaging.

[33]  Noel A Cressie,et al.  Modeling Poisson variables with positive spatial dependence , 1997 .

[34]  Jason Fine,et al.  Estimating equations for association structures , 2004, Statistics in medicine.

[35]  S. Zeger,et al.  Longitudinal data analysis using generalized linear models , 1986 .

[36]  G. Carl,et al.  Analyzing spatial autocorrelation in species distributions using Gaussian and logit models , 2007 .

[37]  Hong-Ye Gao,et al.  Applied wavelet analysis with S-plus , 1996 .

[38]  Daniel A Griffith,et al.  Spatial modeling in ecology: the flexibility of eigenfunction spatial analyses. , 2006, Ecology.

[39]  Michael S. Rosenberg,et al.  Conceptual and Mathematical Relationships among Methods for Spatial Analysis , 2022 .