ggsurface: Plot Model Surface with ggplot2

In wilsonfrantine/R4eco: Set of functions to help ecologists

Plot Model Surface with ggplot2

Description

a ggplot2 2D surface plot function for R glm or lm models

Usage

ggsurface(m, x.var, y.var, x.label, y.label, legend.title,
low.col, high.col, xgrid.size, ygrid.size, n.bins, round.legend, scale.type)

Arguments

m	an object glm, lm or loess object model from stats package
x.var	the name or columns position of the variable to be plot at the x axis. This param has to match with the name or column position of the variable as input in the model object
y.var	same as x.var but for the y axis
x.label	a string with label for the x axis (optional). If NULL the function return x.var
y.label	same as x.label for the y axis
legend.title	a string with name for the response variable. If null the function returns 'response variable'.
low.col	a hexadecimal or ggplot standard color name to represent the low fitting. Default is "grey80"
high.col	a hexadecimal or ggplot standard color name to represent the high fitting. Default is "blue"
xgrid.size	An integer representing the size of the grid to create a prediction. This variable controls the number of elements to build a new x.var interval to fit the model. The default is 15.
ygrid.size	same as xgrid.size for the y variable. The default is also 15.
n.bins	An integer. The number of division / breaks for the plot. More bins will result in a plot with more divisions.
round.legend	the number of decimal cases to round the scales at the legend. The default is 0, which rounds to the next integer. Users may also uses -1 to round for the nearest 10th number.
scale.type	the type of prediction required. The default is "link", and returns the scale of the linear predictors. The alternative "response" return the prediction at the reponse variable scale. So if a glm is ran with a binomial link function than the prediction are probabilities at logit scale, but if type = "response" the prediction will be returned at probabiliti scales. See more in stats::predict help.

Details

This function takes a model object and uses the function prediction to fit the model to a grid of x size. Therefore at this version the function needs one of the model objects from the stats package. In case of any crash, plese contact-me at: <[e-mail]>.

Author(s)

Wilson Frantine-Silva

See Also

stats::predict, stats::glm, stats::lm, stats::lm

Examples

data(mtcars)
m <- glm(mpg ~ wt + hp, data=mtcars, family = "gaussian")
ggsurface(m, x.var = "wt", y.var = "hp",
   legend.title = "milles per galon", high.col = "darkred",
   round.legend = 0)

wilsonfrantine/R4eco documentation built on Jan. 30, 2024, 4:55 p.m.