1

I have two different datasets in one scatter plot, prediction2020_bounds needs to be blue in the legend and median2020 needs to be red in the legend. Here my code:

  ggplot() + 
      
    # Points without corona
    geom_line(data=prediction2020_bounds, aes(x = Date, y = lwr), colour = "blue", linetype = "solid") +
    geom_line(data=prediction2020_bounds, aes(x = Date, y = upr), colour = "blue", linetype = "solid") +
    geom_ribbon(data=prediction2020_bounds, aes(x = Date, ymin = lwr, ymax = upr), alpha=.4) +
    labs(x = "Date", y = "Median Daily Price", title = "Daily Median Price", colour = "Legend Title\n") +
    scale_x_date(date_breaks="1 month", date_labels="%Y %m") +
    scale_colour_manual(labels = c("prediction2020_bounds", "median2020"), values = c(prediction2020_bounds, median2020)) +
    theme_bw() +
    geom_point(data=prediction2020_bounds, aes(x=Date, y=Price),fill="blue",
          colour="darkblue", size=1) +
    
    
    #Points with corona 2020 year
    geom_point(data=median2020, aes(x=Date, y=Price), fill="red",
            colour="red", size=1)
Phil
  • 7,287
  • 3
  • 36
  • 66
Diego Martinez
  • 31
  • 2
  • 1
    It's because you don't assign colours in the `aes()` portion of a layer. If you do `aes(..., colour = "my legend item")`, it will map the colours to a scale, which will generate a legend. – teunbrand Apr 06 '21 at 22:28

1 Answers1

0

Next time please consider include some reproducible example so it would be easier for other to support you in finding a good answer ;)

For legend to be displayed in ggplot2 you need to map the colour or fill inside the mapping param with aes() instead of what you are doing in the questions. The code below is a way to achieve what you want to do.

I also re-format the code a bit in my personal preference where the configuration of the plot scales, labels, theme are all at the bottom while the geom is at the top of ggplot command. This is just personal preference.

Hope this help you get better understanding of the legend in ggplot2.

color_names <- c("prediction 2020 bounds", "median 2020")
color_values <- c("blue", "red")
names(color_values) <- color_names
ggplot() + 
  
  # Points without corona
  geom_line(data=prediction2020_bounds, aes(x = Date, y = lwr,
    # assign the colour to the same name as blue value of color_values
    colour = "prediction 2020 bounds"),  linetype = "solid") +
  geom_line(data=prediction2020_bounds, aes(x = Date, y = upr,
    # assign the colour to the same name as blue value of color_values
    colour = "prediction 2020 bounds"), linetype = "solid") +
  geom_ribbon(data=prediction2020_bounds, aes(x = Date, ymin = lwr, ymax = upr),
    alpha=.4) +
  geom_point(data=prediction2020_bounds, aes(x=Date, y=Price,
    # assign the fill to the same name as blue value of color_values
    fill="prediction 2020 bounds"), 
    colour="darkblue", size=1) +

  #Points with corona 2020 year
  geom_point(data=median2020, aes(x=Date, y=Price, 
    # assign the colour to the same name as red value of color_values
    fill="median 2020"),
    colour="median 2020", size=1) +
  
  # plot configuration scales, theme, etc...
  labs(x = "Date", y = "Median Daily Price",
    title = "Daily Median Price", colour = "Legend Title\n") +
  scale_x_date(date_breaks="1 month", date_labels="%Y %m") +
  scale_colour_manual(values = color_values) +
  scale_fill_manual(values = color_values) +
  theme_bw()

In addition to that you can use theme(legend.position = [position] where [position] can be

  • none: for no legend display
  • left/right/botom/top: corresponded position relate to the main plot area

I found this article is very informative with thorough explaination https://aosmith.rbind.io/2018/07/19/manual-legends-ggplot2/

Sinh Nguyen
  • 4,277
  • 3
  • 18
  • 26
  • Thank you for your answer, it's very close to what I want to have. However, the dots in the legend are both red instead of blue and red. – Diego Martinez Apr 07 '21 at 11:42
  • It really hard to do it without a reproducible example to ensure that we are running on the samething. If you can provide a minimal reproducible example as the link below it would be a lot easier for other to support you. https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example – Sinh Nguyen Apr 07 '21 at 12:25
  • yes thank you for your comment, I will share a new question with a reproducible example. – Diego Martinez Apr 07 '21 at 13:53
  • Please check my new question, this question is broken down to the problem I have – Diego Martinez Apr 07 '21 at 14:08