ggplot2中两个离散比例的图例

Question

I have data which I want to plot along two dimensions: VK (discrete) and relPrec (continous)

Additionally, I have attributes along two dimensions:

type <- c("relPrecAir", "relPrecInv", "relPrectotal")
p <- c(0.9, 0.95, 0.99)

I want to represent type with shapes and p with colors. I manage to do so with the following code:

precisionTable <- structure(list(relPrecTotal = c(0, 0, 0, 0.0070300473755567, 0.00842647198124317, 0.0112374693913644), relPrecAir = c(0, 0, 0, 0.00700370093237813, 0.00839489217055491, 0.011195354831821), relPrecInv = c(0, 0, 0, 0.0108917034179065, 0.0130551941937278, 0.0174102928813975), p = c(0.9, 0.95, 0.99, 0.9, 0.95, 0.99), VK = c(0, 0, 0, 1.15, 1.15, 1.15), case = c("demandStochastic", "demandStochastic", "demandStochastic", "demandStochastic", "demandStochastic", "demandStochastic")), .Names = c("relPrecTotal", "relPrecAir", "relPrecInv", "p", "VK", "case"), row.names = c(NA, -6L), class = "data.frame")

ggplot(precisionTable, aes(VK, relPrecTotal, color=as.factor(p))) + geom_line() + geom_point(shape=19) + 
                    geom_line(aes(VK,relPrecAir, color=as.factor(p))) + geom_point(aes(VK,relPrecAir), shape=17) + 
                    geom_line(aes(VK,relPrecInv, color=as.factor(p))) + geom_point(aes(VK,relPrecInv), shape=15) + 
                        scale_color_discrete(name="p")

One legend is shown for p . But I am not able to draw a second legend for type with the shape values. How can I add it ?

Answer 1

We can work on it by adding new grouping for the parameters relPrecTotal , relPrecAir and relPrecInv , then proceed to plot it using ggplot.

EDIT Overriding the shape in color's legend

library(reshape2)
precisionTable2 <- melt(precisionTable, measure.vars = c("relPrecTotal", "relPrecAir", "relPrecInv"))
ggplot(precisionTable2, aes(VK, value)) +
  geom_line(aes(color=as.factor(p), shape=as.factor(variable))) +
  geom_point(aes(color=as.factor(p),shape=as.factor(variable))) +
  scale_color_discrete(name='p') +
  scale_shape_discrete(name='relPrec') +
  guides(color=guide_legend(override.aes = list(shape=NA)))

Answer 2

You could melt your data.frame to avoid the multiple steps in your ggplot command, and create a new variable (here ptShape ) that you would map within the aes of a single geom_point call:

library(reshape2)
pmelt = melt(precisionTable, id.vars=c('VK', 'p', 'case'))
pmelt$ptShape = rep(c(19,17,15), each=6)

ggplot(pmelt, aes(x=VK, y=value)) + 
  geom_line(aes(colour=as.factor(p), shape=as.factor(ptShape))) +
  geom_point(aes(colour=as.factor(p), shape=as.factor(ptShape)), size=5)