mgcv provides functions for generalized additive modelling (gam and bam) and
generalized additive mixed modelling (gamm, and random.effects). The term GAM is taken to include
any model dependent on unknown smooth functions of predictors and estimated by quadratically penalized (possibly quasi-) likelihood maximization. Available distributions are covered in family.mgcv and available smooths in smooth.terms.
Particular features of the package are facilities for automatic smoothness selection (Wood, 2004, 2011), and the provision of a variety of smooths of more than one variable. User defined smooths can be added. A Bayesian approach to confidence/credible interval calculation is provided. Linear functionals of smooths, penalization of parametric model terms and linkage of smoothing parameters are all supported. Lower level routines for generalized ridge regression and penalized linearly constrained least squares are also available.
mgcv provides generalized additive modelling functions gam,
predict.gam and plot.gam, which are very similar
in use to the S functions of the same name designed by Trevor Hastie (with some extensions).
However the underlying representation and estimation of the models is based on a
penalized regression spline approach, with automatic smoothness selection. A
number of other functions such as summary.gam and anova.gam
are also provided, for extracting information from a fitted gamObject.
Use of gam is much like use of glm, except that
within a gam model formula, isotropic smooths of any number of predictors can be specified using
s terms, while scale invariant smooths of any number of
predictors can be specified using te, ti or t2 terms.
smooth.terms provides an
overview of the built in smooth classes, and random.effects should be refered to for an overview
of random effects terms (see also mrf for Markov random fields). Estimation is by
penalized likelihood or quasi-likelihood maximization, with smoothness
selection by GCV, GACV, gAIC/UBRE or (RE)ML. See gam, gam.models,
linear.functional.terms and gam.selection for some discussion of model specification and
selection. For detailed control of fitting see gam.convergence,
gam arguments method and optimizer and gam.control. For checking and
visualization see gam.check, choose.k, vis.gam and plot.gam.
While a number of types of smoother are built into the package, it is also
extendable with user defined smooths, see smooth.construct, for example.
A Bayesian approach to smooth modelling is used to derive standard errors on
predictions, and hence credible intervals (see Marra and Wood, 2012). The Bayesian covariance matrix for
the model coefficients is returned in Vp of the
gamObject. See predict.gam for examples of how
this can be used to obtain credible regions for any quantity derived from the
fitted model, either directly, or by direct simulation from the posterior
distribution of the model coefficients. Approximate p-values can also be obtained for testing
individual smooth terms for equality to the zero function, using similar ideas (see Wood, 2013a,b). Frequentist
approximations can be used for hypothesis testing based model comparison. See anova.gam and
summary.gam for more on hypothesis testing.
For large datasets (that is large n) see bam which is a version of gam with
a much reduced memory footprint.
The package also provides a generalized additive mixed modelling function,
gamm, based on a PQL approach and
lme from the nlme library (for an lme4 based version, see package gamm4).
gamm is particularly useful
for modelling correlated data (i.e. where a simple independence model for the
residual variation is inappropriate). In addition, low level routine magic
can fit models to data with a known correlation structure.
Some underlying GAM fitting methods are available as low level fitting
functions: see magic. But there is little functionality
that can not be more conventiently accessed via gam .
Penalized weighted least squares with linear equality and inequality constraints is provided by
pcls.
For a complete list of functions type library(help=mgcv). See also mgcv.FAQ.
These provide details for the underlying mgcv methods, and fuller references to the large literature on which the methods are based.
Wood, S.N., N. Pya and B. Saefken (2016), Smoothing parameter and model selection for general smooth models (with discussion). Journal of the American Statistical Association 111, 1548-1575 http://dx.doi.org/10.1080/01621459.2016.1180986
Wood, S.N. (2011) Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society (B) 73(1):3-36
Wood, S.N. (2004) Stable and efficient multiple smoothing parameter estimation for generalized additive models. J. Amer. Statist. Ass. 99:673-686.
Marra, G and S.N. Wood (2012) Coverage Properties of Confidence Intervals for Generalized Additive Model Components. Scandinavian Journal of Statistics, 39(1), 53-74.
Wood, S.N. (2013a) A simple test for random effects in regression models. Biometrika 100:1005-1010
Wood, S.N. (2013b) On p-values for smooth components of an extended generalized additive model. Biometrika 100:221-228
Wood, S.N. (2017) Generalized Additive Models: an introduction with R (2nd edition), CRC
Development of mgcv version 1.8 was part funded by EPSRC grants EP/K005251/1 and EP/I000917/1.
# NOT RUN {
## see examples for gam and gamm
# }
Run the code above in your browser using DataLab