ggplot multi-factor-level grouping for boxplot wit

问题:

I'm trying to create a boxplot of the following data

Temp<-rnorm(90,mean=100,sd=10)
Yr<-sample(c("1999","2000","2005","2009","2010"),size=90,replace=TRUE)
Month<-sample(c("June","July","August"),size=90,replace=TRUE)
Month
df<-data.frame(Temp,Month,Yr)

The visual I want and its corresponding code are below:

ggplot(df,aes(x=interaction(Month,Yr),y=Temp,fill=Month))+
    geom_boxplot()+
    xlab("Year")+
    ylab("Daily Maximum Temperature")

You'll notice, though, that there are a few years missing from the data, and I'm trying to make the plot reflect that with gaps in the x-scale. The other problem is the text and tick marks on the axis. I'd like the ticks to just be the Year of observation rather than Month.Year since the month is already coded in the fill. I've tried scale_x_discrete, but trying to supply discrete values for a continuous axis spits out a blank graph and an error. I've met my swearing at the computer quota for the day, and it would be really awesome to get a little help on this.

回答1:

This creates huge gaps, as every year gets its own gap, but you can adapt this by passing only specific years as the levels argument to the factor() call.

df$Yr <- factor(df$Yr, levels=1999:2010)

ggplot(df,aes(x=Yr,y=Temp,fill=Month))+
  geom_boxplot(position=position_dodge(1))+
  ylab("Daily Maximum Temperature") +
  scale_x_discrete("Year", drop=FALSE)