Ggplot using group_by

Let's say I measured some values ​​in two experiments, an example of a toy in R:

set.seed(9) 
df <- data.frame(
    exp=c(rep(1,10), rep(2,10)),
    value=runif(20,0,3))

      

        
        
        
      

    

Then I assign a categorical variable level

based on value

:

require("dplyr") 
df <- df %>% mutate(level = ifelse(value<1, "low", ifelse(value>2, "high", "intermediate")))  

      

        
        
        
      

    

I can display geom_point()

geom ( ggplot2

package) value

for two groups exp

:

require("ggplot2")
ggplot(df, aes(x=factor(exp), y=value))+geom_point()

      

        
        
        
      

    

My problem: How can I break the display using facet_wrap()

on level

to get a 3x3 plot geom_point()

with nine combinations level

for two exp

(for example, "high", "medium" and "low" vertically for experiment 1 and horizontally for experiment 2. In other words, the top left graph represents those value

that are "high" in experiments 1 and 2, the first column of the first row is those that are "high" in experiment 1 and "intermediate" in experiment 2, etc.). Can I somehow use group_by()

the call ggplot2, or do I need to do variable level.exp1

and level.exp2

for the face?

UPDATE: I don't seem to explain this properly, so to clarify what I need: I would like to facet_wrap()

instead repeat this code 9 times with different conditions filter()

(i.e. For all 9 combinations level

in experiments 1 and 2)

df  %>% filter((exp==1 & level=="high") | (exp==2 & level=="high")) %>%
  ggplot(aes(x=factor(exp), y=value))+geom_point()

      

        
        
        
      

    

Handmade makeup and unrelated to the df

above and fake legends can give an idea of ​​what I want:

There must be a gentle solution - I couldn't find a solution to a dcast()

.