cumimp.combinedGradientForest: Cumulative Importance of predictors for a...

In gradientForest: Random Forest functions for the Census of Marine Life synthesis project.

Cumulative Importance of predictors for a combinedGradientForest object

Description

Provide access to the cumulative importance values of predictors after combining gradientForest objects.

Usage

## S3 method for class 'combinedGradientForest'
cumimp(x, predictor, weight = c("uniform", "species", "rsq.total", 
"rsq.mean", "site", "site.species", "site.rsq.total", "site.rsq.mean")[3], gf.name, ...)

Arguments

x	an object of class combinedGradientForest generated by combinedGradientForest.
predictor	the name of the predictor variable.
weight	type of weighting to perform across gradientForest objects. The choices are uniform: each gradientForest object gets weight 1 species: each gradientForest object gets weight equal to the number of species rsq.total: each gradientForest object gets weight equal to the total R^2 (default) rsq.mean: each gradientForest object gets weight equal to the mean R^2 per species site: each gradientForest object gets weight equal to the number of sites site.species: each gradientForest object gets weight equal to the number of species times the number of sites site.rsq.total: each gradientForest object gets weight equal to the total R^2 times the number of sites site.rsq.mean: each gradientForest object gets weight equal to the mean R^2 per species times the number of sites
gf.name	name of gradientForest object required. If supplied, provide the cumulative importance for the specific gradientForest object. weight is ignored in this case.
...	further arguments passed to or from other methods.

Details

Cumulative importance curves are found for each predictor by weighting across the gradientForest objects. See the section Compositional turnover and equation (9) in Ellis et al (2012).

Value

If gf.name is missing, a list with components x and y, where x is the predictor values and y is the combined cumulative importance using the specified weighting across gradientForest objects. If gf.name is supplied, the list for the predictor from the specified gradientForest object is returned.

Author(s)

N. Ellis, CSIRO, Cleveland, Australia. <[e-mail]>. S.J. Smith, DFO, Dartmouth, NS, Canada. <[e-mail]>

References

Ellis, N., Smith, S.J., and Pitcher, C.R. (2012) Gradient Forests: calculating importance gradients on physical predictors. Ecology, 93, 156–168.

See Also

combinedGradientForest

Examples

data(CoMLsimulation)
preds <- colnames(Xsimulation)
specs <- colnames(Ysimulation)
f1 <- gradientForest(data.frame(Ysimulation,Xsimulation), preds, specs[1:6], ntree=10)
f2 <- gradientForest(data.frame(Ysimulation,Xsimulation), preds, specs[1:6+6], ntree=10)
f12 <- combinedGradientForest(west=f1,east=f2)
par(mfrow=c(1,3))
plot(cumimp(f12, "A", gf.name="west"),main="West")
plot(cumimp(f12, "A", gf.name="east"),main="East")
plot(cumimp(f12, "A", weight = "rsq.total"),main="Combined")

gradientForest documentation built on Aug. 24, 2023, 3:03 p.m.