CODEWITHSUNDEEP
rggplot2
ahj
ahj

Reputation: 15

Factor level/color present in data but missing from plot using scale_color_manual

Here is some dummy data:

dummy <- c(rep("castor", 20), rep("dandelion", 20), rep("goose", 20), rep("teapot", 20), rep("bee", 20), rep("fan" , 20))
groups <- c(rep(c(rep(1, 10), rep(2, 10)), 6))
mydata <- data.frame(a = rnorm(120, 0, 1), b = rnorm(120, 0, 1), dummy = dummy, groups = groups)

Here is the function I wrote to plot a level of my choosing from the dummy column in blue. I want the remaining levels to be plotted in gray. Those remaining factor levels (let's call them reference) are hardcoded now which is not a problem because in my real data I have more levels of my choice (let's call them test) to plot, which is why I am writing a function for this. 

custom_plot <- function(level){
   df <- subset(mydata, mydata$dummy == level | mydata$groups == "1")

   # Check that the test level is indeed in the subset df
   return(table(df$dummy, df$groups))              

   p <- ggplot(df, aes(x = a, y = b, color = dummy, shape = dummy)) + geom_point() + theme_bw()
   p <- p + scale_color_manual(values = c(level = "blue", "castor" = "gray", "dandelion" = "gray", "fan" = "gray", "goose" = "gray", "teapot" = "gray"))
   p
}

custom_plot("bee")

I can't post an image because I need at least 10 reputation, but essentially the graph I get has the test level in the legend, but the actual datapoints are missing from the graph, and the shape/color of that level is blank. I tried using the breaks, limits, and labels arguments but nothing worked.

Thank you for your help!

Upvotes: 0

Views: 436

Answers (1)

cdeterman
cdeterman

Reputation: 19960

I believe this should provide what you are looking for. It is typically much easier if you just pass the list of colors to ggplot. I took the liberty of showing you a way to index the colors so that the dummy level you choose will be the one that is colored blue.

library(ggplot2)
custom_plot <- function(level){
  df <- subset(mydata, mydata$dummy == level | mydata$groups == "1")

  # create colors vector
  idx <- which(levels(mydata$dummy) == level)
  nlev <- nlevels(mydata$dummy)
  colors <- c(rep("gray", idx-1), "blue", rep("gray", nlev-idx))

  p <- ggplot(df, aes(x = a, y = b, color = dummy, shape = dummy)) + 
    geom_point() + 
    theme_bw()
  p <- p + scale_color_manual(values = colors)
  p
}

custom_plot("bee")

Upvotes: 1

Related Questions