Although inference is a critical component in ecological modeling, the balance between accurate predictions and inference is the ultimate goal in ecological studies (Peters 1991; De'ath 2007). Practical applications of ecology in conservation planning, ecosystem assessment, and bio-diversity are highly dependent on very accurate spatial predictions of ecological process and spatial patterns (Millar et al. 2007). However, the complex nature of ecological systems hinders our ability to generate accurate models using the traditional frequentist data model (Breiman 2001a; Austin 2007). Well-defined issues in ecological modeling, such as complex non-linear interactions, spatial autocorrelation, high-dimensionality, non-stationary, historic signal, anisotropy, and scale contribute to problems that the frequentist data model has difficulty addressing (Olden et al. 2008). When one critically evaluates data used in ecological models, rarely do the data meet assumptions of independence, homoscedasticity, and multivariate normality (Breiman 2001a). This has caused constant reevaluation of modeling approaches and the effects of reoccurring issues such as spatial autocorrelation. © 2011 Springer Science+Business Media, LLC.
CITATION STYLE
Evans, J. S., Murphy, M. A., Holden, Z. A., & Cushman, S. A. (2011). Modeling species distribution and change using random forest. In Predictive Species and Habitat Modeling in Landscape Ecology: Concepts and Applications (pp. 139–159). Springer New York. https://doi.org/10.1007/978-1-4419-7390-0_8
Mendeley helps you to discover research relevant for your work.