Link zu WSL Hauptseite Eidg. Forschungsanstalt für Wald, Schnee und Landschaft WSL
 

ESMs - Ensembles of Small Models for R

What are ESMs?

Ensembles of Small Models (ESMs) are simple small models, e.g. bivariate models of only two predictors, which are averaged to an ensemble weighted by the mean of a cross-validated evaluation score (e.g. AUC).

What is the advantage of ESMs?

In Species Distribution Modeling it is often difficult to model rare species because only few occurrences are available and models are prone to overfitting. ESMs avoid overfitting without losing explanatory power through variable reduction by averaging small models. ESMs are therefore especially advantageous for modeling rare species.

How to start?

Here, we supply a commented R script to build ESMs in R. ESM functions are provided as part of the R package ecospat (Broenniman et al. 2016).

Further information?

Please have also a look at our blog post for methods in ecology and evolution. We would appreciate it very much if you leave a comment.

ESMs were granted with the MCED-award 2015.

Download

ESM functions are available in the R package ecospat (Broenniman et al. 2016).

Here we provide an example using original data from Breiner et al. (2015) and show how ESMs could be build step-by-step:

ESM-scripts.zip

References

Breiner F.T.; Guisan A.; Bergamini A.; Nobis M.P., 2015:

Overcoming limitations of modelling rare species by using ensembles of small models. Methods Ecol. Evol. 6: 1210-1218.

Broenniman O.; Di Cola V.; Guisan A., 2016:
ecospat: Spatial Ecology Miscellaneous Methods. R package version 2.0. Available from World Wide Web <http://cran.r-project.org/package=ecospat>

Lomba, A.; Pellissier, L.; Randin, C.F.; Vicente, J.; Moreira, F.; Honrado, J.; Guisan A., 2010:

Overcoming the rare species modelling paradox: A novel hierarchical framework applied to an Iberian endemic plant. Biological Conservation, 143, 2647–2657.

 

Spacer