How to exclude facet wrap/grid with less than n number of observations in R-CodePudding

I am using ggplot to create numerous dot charts (using geom_point and facet_wrap) for two variables with multiple categories. Multiple categories have only a few points, which I cannot use,

Is there a way to set a minimum number of observations for each facet (i.e. only show plots with 10 or more observations)?

by_AB <- group_by(df, A, B)

by_AB%>% 
  ggplot(aes(X,Y))  
  geom_point()  
  facet_wrap(A~B,  scales="free")   
  geom_smooth(se = FALSE)  
  theme_bw()```

CodePudding user response：

It is best to remove the small groups from your data before plotting. This is very easy if you do

df %>% 
  group_by(A, B) %>% 
  filter(n() > 1) %>%
  ggplot(aes(X,Y))  
  geom_point()  
  facet_wrap(A~B,  scales="free")   
  geom_smooth(se = FALSE)  
  theme_bw()

Obviously, we don't have your data, so here is an example using the built-in mtcars data set. Suppose we want to plot mpg against wt, but facet by carb:

library(tidyverse)

mtcars %>%
  ggplot(aes(wt, mpg))  
  geom_point()  
  facet_wrap(.~carb)

Two of our facets look out of place because they only have a single point. We can simply filter these groups out en route to ggplot:


mtcars %>%
  group_by(carb) %>%
  filter(n() > 1) %>%
  ggplot(aes(wt, mpg))  
  geom_point()  
  facet_wrap(.~carb)

^{Created on 2022-08-25 with reprex v2.0.2}