Ggplot with patterns and colors - legend-CodePudding

I need to make a bar plot by country and sources of income. As I have 13 income components by each single country, I'm using colors and patterns to make a clearer distinction. Here a subsample of the data:

 structure(list(country = c("Austria", "Austria", "Austria", "Austria", 
 "Austria", "Austria", "Austria", "Austria", "Austria", "Austria", 
 "Austria", "Austria", "Belgium", "Belgium", "Belgium", "Belgium", 
 "Belgium", "Belgium", "Belgium", "Belgium", "Belgium", "Belgium", 
 "Belgium", "Belgium", "Germany"), `relative contributions` = c(1.12873709201813, 
  0.190666541457176, 0.0881932526826859, -0.0167537834495306, -0.0258312318474054, 
  0.215001374483109, -0.00148478976916522, -0.000572826422285289, 
  -0.0133617669343948, -0.00855048839002848, -0.00286591215990484, 
  -0.55317747592926, 1.43853068351746, 0.12194737046957, 0.0775384828448296, 
  -0.00403465563431382, -0.0196540541946888, -0.0183424148708582, 
  -0.0311897918581963, -0.00196828553453088, -0.00494809495285153, 
  -0.0225886180996895, -0.000150939886225387, -0.535139620304108, 
  1.23838818073273), component = c("Employee wages", "Self-employment", 
  "Capital", "Private transfers", "Unemployment", "Old-age   survivors", 
 "Sick   disability", "Education", "Social exclusion allowances", 
 "Family allowances", "Housing allowances", "Taxes", "Employee wages", 
 "Self-employment", "Capital", "Private transfers", "Unemployment", 
 "Old-age   survivors", "Sick   disability", "Education", "Social exclusion allowances", 
 "Family allowances", "Housing allowances", "Taxes", "Employee wages"
 )), row.names = c(NA, -25L), class = c("tbl_df", "tbl", "data.frame"
 ))

Up to now, I was able to produce the following plot:

 pat <- c('bricks', 'fishscales', 'verticalsaw', 'crosshatch', 'checkboard', 'octagons', 'verticalbrics', 'vertical', 'right45', 'hs_horizontal', 'horizontalsaw', 'hs_cross', 'hexagons')

ggplot(relative)   
 geom_bar_pattern(aes(x=country, y=`relative contributions`, fill=factor(component), pattern_type=factor(component)), stat="identity", position="stack", width = 0.6, pattern='magick',  pattern_scale = 0.5, colour="black")   
  scale_fill_manual(name="Income components", values = mycolors)    scale_pattern_type_discrete(choices = pat, guide = 'none')   
 scale_y_continuous(limits = c(-1,2), breaks = scales::pretty_breaks(n = 7))   
 theme_bw()  
 theme(axis.text.x = element_text(angle=45, hjust=1, vjust=1))   xlab("")   ylab("")   theme(plot.title = element_text(hjust = 0.5))   `

Which results in the following figure:

However, the legend does not report the pattern. If I offset guide="none" I will have two legends, but is there a way to have a unique legend (as in the current version) reporting both the colors and the patterns?

Many thanks for the help!

CodePudding user response：

Remove guide="none" to show the patterns in the legend but add name="Income components" to scale_pattern_xxx so that the guides get merged. Otherwise the scales have different names and you end up with two separate legends.

Note: I switched to scale_fill_discrete as mycolors was not provided.

library(ggplot2)
library(ggpattern)

ggplot(relative)  
  geom_bar_pattern(aes(x = country, y = `relative contributions`, fill = factor(component), pattern_type = factor(component)), stat = "identity", position = "stack", width = 0.6, pattern = "magick", 
                   pattern_scale = 0.5, colour = "black")  
  scale_fill_discrete(name = "Income components")  
  scale_pattern_type_discrete(name = "Income components", choices = pat)  
  scale_y_continuous(limits = c(-1, 2), breaks = scales::pretty_breaks(n = 7))  
  theme_bw()  
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))  
  xlab("")  
  ylab("")  
  theme(plot.title = element_text(hjust = 0.5))