如何在满足标准的某些群体中绘制R图？

Question

Here's a simple dataset I made:

test <- data.frame(
   ID=1:10,
   group=c(3,2,1,3,3,3,4,1,3,3),
   height=c(156,167,165,187,153,172,178,191,155,189)
)

And a look at the how many individuals are in each group:

> table(test$group)
1 2 3 4 
2 1 6 1 

Then I did a boxplot

boxplot(test$height~test$group)

As you can see group 2 and group 4 only have one individual in them.. I want to exclude them when doing the boxplot. In other words, do the plot on groups that have more than one observation?

I know the subset function, and think this might be useful, but not sure how to best apply it in this case.

Answer 1

You can use transform() and ave() to add a column indicating how many observations are in each group and then use the subset() parameter to only keep those with more than 1 obs. For example

boxplot(height~group, 
    transform(test, groupcount=ave(ID, group, FUN=length)), 
    subset=groupcount>1)

Note that you can only use the subset= parameter of boxplot() when you use the formula syntax.

Answer 2

I like dplyr for these sorts of things

library(dplyr)
test %>% group_by(group) %>%
         filter(n() > 1) %>% 
         boxplot(height~group, .) 

Answer 3

You can do this:

tab <- as.data.frame(table(test$group))
tab1 <- tab[which(tab$Freq > 1),]
test2 <- test[which(test$group %in% tab1$Var1),]
boxplot(test2$height~test2$group)