WollMilchSauPlot: Compare means, distributions, and correlations of different...

In ModelDataComp: Model-Data Comparison

Description

Plots a pirateplot with couulors according to model performance. The function can be used to compare means, distributions, and correlations (or any other metric from ObjFct) between different datasets and models. Thus this plot is almost an "eier-legende Wollmichsau" (german, animal that produces eggs, wool, milk and meet) for model-data comparison. The plot is based on the function pirateplot

Usage

WollMilchSauPlot(x, ref = 1, objfct = "Cor", cols = NULL, brks = NULL, 
    names = NULL, main = NULL, xlab = NULL, ylab = NULL, xlim = NULL, 
    ylim = NULL, legend = TRUE, legend.only = FALSE, cut.min = NULL, 
    cut.max = NULL, avg = TRUE, points = TRUE, bean = TRUE, inf = TRUE, 
    bar = TRUE, ...)

Arguments

x	a data.frame with at least two columns
ref	Which column in x is the reference dataset? Also more than one reference can be provided, e.g. ref = c(1,2) will compute the objfct based on the combination of both datasets.
objfct	Which objective function metric should be used to create the colour palette? (see ObjFct)
cols	vector of colors from which the color palette should be interpolated
brks	break for colour scale
names	names of the datasets
main	title of the plot
xlab	label for x-axis
ylab	label for y-axis
xlim	limits for x-axis
ylim	limits for y-axis
legend	plot a legend?
legend.only	plot only a legend?
cut.min	Optional minimum value of the beans.
cut.max	Optional maximum value of the beans.
avg	plot average line?
points	plot points?
bean	plot beans (density estimates)?
inf	plot inference bands around mean?
bar	plot bars?
...	further arguments to pirateplot

Details

No details.

Author(s)

Matthias Forkel <[e-mail]> [aut, cre]

References

No reference.

See Also

pirateplot, ObjFct, TaylorPlot, WollMilchSauPlot, ScatterPlot

Examples

# create some data
obs <- rlnorm(500, 1, 1) # observations
sim1 <- obs * rnorm(500, 1, 0.5) # similar to obs
sim2 <- obs * rnorm(500, 1, 2) # less similar to obs
sim3 <- obs * rnorm(500, 1, 4) # less similar to obs
sim4 <- rlnorm(500, 1, 1) # same distribution but no correlation
sim5 <- rnorm(500, 4.4, 2) # similar mean but different distribution
x <- data.frame(obs, sim1, sim2, sim3, sim4, sim5)
x[x < 0] <- 0

# default plot
WollMilchSauPlot(x)

# with different objective function as colour
WollMilchSauPlot(x, objfct="IoA")
WollMilchSauPlot(x, objfct="Pbias")
WollMilchSauPlot(x, objfct="FV")

# axis labels and title
WollMilchSauPlot(x, ylab="Area (km2)", xlab="Groups", main="Comparison")

# remove certain elements from plot
WollMilchSauPlot(x, points=FALSE)
WollMilchSauPlot(x, bean=FALSE)
WollMilchSauPlot(x, points=FALSE, bean=FALSE)
WollMilchSauPlot(x, points=FALSE, bean=FALSE, bar=FALSE)
WollMilchSauPlot(x, inf=FALSE)
WollMilchSauPlot(x, inf=FALSE, avg=FALSE)
WollMilchSauPlot(x, avg=FALSE, bar=FALSE, inf=FALSE)

# different color palettes
WollMilchSauPlot(x, cols=c("blue", "red"))
WollMilchSauPlot(x, cols=c("blue", "grey", "red"))
WollMilchSauPlot(x, cols=rainbow(10))
WollMilchSauPlot(x, objfct="IoA", cols=heat.colors(5))
WollMilchSauPlot(x, objfct="RMSE", cols=rev(heat.colors(5)))

# without legend (but using an objective function to colour)
WollMilchSauPlot(x, legend=FALSE)

# only legend
WollMilchSauPlot(x, legend.only=TRUE)

# without using an objective function - categorial colours
WollMilchSauPlot(x, objfct=NULL)

# different example data
obs <- rnorm(500, 5, 1)
sim1 <- obs * rnorm(500, 1, 0.2) # similar to obs
sim2 <- obs * rnorm(500, 2, 1) # bias
sim3 <- obs * rlnorm(500, 1, 0.1) # less similar to obs but highly correlated
x <- data.frame(obs, sim1, sim2, sim3)
WollMilchSauPlot(x)

ModelDataComp documentation built on Nov. 22, 2020, 3 a.m.