Geom_bar with position dodge on one categorical variable and colour on another

Question

I'd like to create a bar plot with dodged bars based on activity and the bar colour dictated by thres or alternatively conditionally: cpue > 75 = 'green' cpue < 75 & cpue > 50 = 'orange' cpue < 50 = 'purple'

An example of the data is:

     week area_code activity  cpue  p_legal  thres thres2 
   <dbl>     <chr>    <chr>    <dbl>    <dbl>  <chr>  <chr>  
1    30      2L12  Looking      NaN      NaN  green  green  
2    31      2L12  Fishing 81.96667 90.00000  green  green  
3    31      2L12  Looking 61.32353 90.00000 orange  green 
4    32      2L12  Fishing 87.07883 80.00000  green  green  
5    32      2L12  Looking 66.99622 94.33333 orange  green 
6    35      2L12  Fishing 60.29923 86.25000 orange  green 

I know I can use a facet wrap like:

ggplot(plot_df[plot_df$area_code == area, ]) +
  geom_bar(aes(x=week, y=cpue, fill=thres), stat='identity') +
  scale_fill_manual(values=c('green'= 'green', 'orange'='orange', 'purple' = 'purple'), guide=FALSE) +
  geom_hline(yintercept = c(50, 75)) +
  facet_wrap(~activity)

What I'd actually like is dodged bars but the problem I'm encountering is that geom bar dodges bars based on the fill aesthetic and the colour aesthetic only changes the border colour e.g.

 ggplot(plot_df[plot_df$area_code == area, ]) +
  geom_bar(aes(x=week, y=cpue, fill=activity, colour = thres), position = 'dodge', stat='identity') +
  geom_hline(yintercept = c(50, 75))

I'd like the same output as above but bar colour dictated by the condition outlined or thres

Answer 1

Facets is the preferred solution for additional factors, when you have already used fill for a factor. If you want to fill by two factors, you could try using interaction():

ggplot(plot_df[plot_df$area_code == area, ]) +
  geom_col(aes(week, cpue, fill = interaction(activity, thres)), position = "dodge")