GAMs with integrated model selection using penalized regression splines and applications to environmental modelling

By Simon N. Wood, Nicole H. Augustin. 2002

821 citations recorded by the source corpus.

Subfields: Electrical and Electronic Engineering, Environmental Engineering, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Electrical and Electronic Engineering

In Environmental Engineering

In Statistics and Probability

Citations in the corpus

Invariant Causal Prediction for Sequential Data (2017, methodology): “For example, for very smooth shifts one could use generalized additive models (GAM; Wood and Augustin 2002), implemented in the R-package mgcv, to get the nonlinear smoothing fit and then consider a measure of how far the smoother deviates from the horizontal line at 0.”
A field test of the stress-gradient hypothesis along an aridity gradient (2011, methodology): “…relationship – linear or non-linear; i.e. this nonparametric category of models allows determination of the shape of the response curves from the data instead of fitting an a priori parametric model that is limited in its available shape of response (Wood & Augustin 2002; Caprio et al. 2009).”
Predicting tree species presence and basal area in Utah: A comparison of stochastic gradient boosting, generalized additive models, and tree-based methods (2006, methodology): “Recent advances in GAM methodology have been made by Wood and Augustin (2002), as well as Yee and MacKenzie (2002).”
A generalized additive model for the spatial distribution of snowpack in the Spanish Pyrenees (2005, methodology): “Thus, some studies to map snowpack have been based on the linear relationships between snow depth and topographical and locational variables (Caine, 1975; Chang and Li, 2000), but the non-linearity of the relationship between the variance of snowpack and terrain characteristics reduces the capacity…”
Diel vertical migration in the Lake Superior pelagic community. I. Changes in vertical migration of coregonids in response to varying predation risk (2006, methodology): “Penalized regression splines allow the flexibility of the curve to be optimized by “rewarding” model fit and “penalizing” model degrees of freedom (Wood and Augustin 2002) rather than setting the model degrees of freedom a priori.”
Comparison of five modelling techniques to predict the spatial distribution and abundance of seabirds (2012, methodology): “In GAMs, we used the automatic term selection procedure which imposes a penalty to smooth functions and can thus effectively remove terms from the model (Wood and Augustin, 2002).”
ON CONFIDENCE INTERVALS FOR GENERALIZED ADDITIVE MODELS BASED ON PENALIZED REGRESSION SPLINES (2006, methodology): “…Association Inc. intervals in non-Gaussian settings, in particular for GAMs constructed from penalized regression splines (e.g. Marx & Eilers, 1998; Wood & Augustin, 2002; Wood, 2003); (iv) provides quite extensive simulation tests of the intervals and (v) suggests a simple method for dealing…”
Spatial interpolation of monthly climate data for Finland: comparing the performance of kriging and generalized additive models (2013, background): “GAM have the form (Hastie and Tibshirani 1990; Wood and Augustin 2002) g(μ) = β0 + s1(x1) + s2(x2) + ... + sk(xk), (10) where g(μ) is the link function, β0 is a constant, sk is the smoothing parameter to be estimated, and xk is the explanatory variable.”
Modelling potential spawning habitat of sardine (Sardina pilchardus) and anchovy (Engraulis encrasicolus) in the Bay of Biscay (2007, methodology): “The selection of the GAMs smoothing predictors was done following the method proposed by Wood and Augustin (2002), using the 'mgcv' library in the R statistical software (R Development Core Team, 2004).”
Predicting climate‐induced range shifts: model differences and model reliability (2006, methodology): “We used the MGCV package in R to fit all GAM models (Wood & Augustin, 2002).”
Spatial fisheries ecology: Recent progress and future prospects (2008, background): “These include, but are not limited to, distribution study of eggs (Fox et al., 2000; Wood and Augustin, 2002; Ciannelli et al., 2007b), tuna (Zagaglia et al., 2004), seatrout (Kupschus, 2003), crabs (Jensen et al., 2005), squids (Denis et al., 2002), flatfish (Swartzman et al., 1992; Simpson and…”
The big five personality traits and parental burnout: Protective and risk factors (2017, neutral): “It is easier to analyze nonlinear interactions between numerical explanatory variables with GAMs than with polynomial transformations (Wood & Augustin, 2010) in the GLM framework.”
Competitive exclusion along climate gradients: energy efficiency influences the distribution of two salmonid fishes (2011, methodology): “At each step in model selection, variables were added or removed by comparing models using a general cross-validation procedure – the UBRE scores [the lowest score is the best (Wood & Augustin, 2002)] – which allows minimizing the trade-off between model fit and the overall smoothness.”
Winter distribution of blue crab Callinectes sapidus in Chesapeake Bay : application and cross-validation of a two-stage generalized additive model (2005, methodology): “Penalized regression splines (Wood & Augustin 2002) were fitted using the mgcv (v1.”