Default colors in ggplot2 for extended legend labels

Question

I wanted to refer to the question: Force ggplot legend to show all categories when no values are present

I'm in a similar situation, but I would like the colors to be the default.

How should I do this?

ADDED:

I looked more closely and it turned out that, unfortunately, the labels were changed.

Raw data looks like this:

> str(mj)
'data.frame':   4393 obs. of  22 variables:
 $ OS_Gatunek       : Factor w/ 5 levels "Taraxacum ancistrolobum",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ PH_CreateDate    : Factor w/ 15 levels "2016-04-06","2016-04-19",..: 2 2 2 2 2 2 2 2 2 2 ...
 $ L_Ksztalt        : Factor w/ 3 levels "lancetowaty",..: 3 2 3 3 2 2 3 3 2 3 ...
 $ L_Symetria       : Factor w/ 3 levels "duża","mała",..: 1 3 1 3 2 3 2 2 2 1 ...
 $ L_Sfaldowanie    : Factor w/ 2 levels "brak","obecne": 1 1 1 2 2 1 1 2 1 1 ...
 $ KS_Ksz           : Factor w/ 3 levels "hełmiasty","strzałkowaty",..: 2 3 1 1 3 1 1 1 1 1 ...
 $ KS_KszWierz      : Factor w/ 5 levels "spiczasty","tępo spiczasty",..: 3 1 5 2 2 1 1 2 3 4 ...
 $ KS_KszKrGor      : Factor w/ 10 levels "esowaty","odwrotnie esowaty",..: 7 7 10 1 7 10 10 10 10 10 ...
 $ KS_KszKrDol      : Factor w/ 10 levels "esowaty","odwrotnie esowaty",..: 9 7 10 7 7 9 9 10 9 9 ...
 $ KS_Zab           : Factor w/ 2 levels "brak","obecne": 1 1 1 1 1 1 1 1 1 1 ...
 $ KS_TendTworzKlap : Factor w/ 2 levels "brak","obecna": 1 1 1 1 1 1 2 1 1 2 ...
 $ KB_Ustawienie    : Factor w/ 5 levels "odchylone","odgięte",..: 1 1 1 3 1 1 1 1 1 1 ...
 $ KB_Zakonczenie   : Factor w/ 5 levels "ostro spiczaste",..: 3 3 2 3 2 2 5 5 3 2 ...
 $ KB_KsztKrawGornej: Factor w/ 10 levels "esowaty","odwrotnie esowaty",..: 10 1 10 7 7 10 10 10 10 1 ...
 $ KB_KsztKrawDolnej: Factor w/ 10 levels "esowaty","odwrotnie esowaty",..: 9 7 10 7 2 10 9 2 9 1 ...
 $ KB_ZabkKrGornej  : Factor w/ 2 levels "brak","obecne": 2 1 1 1 1 1 1 2 1 1 ...
 $ KB_ZabkKrDolnej  : Factor w/ 2 levels "brak","obecne": 1 2 1 1 1 1 1 1 1 1 ...
 $ KB_TendDoTwKlap  : Factor w/ 2 levels "brak","obecna": 1 1 1 1 1 1 1 1 1 1 ...
 $ I_Ksztalt        : Factor w/ 3 levels "całe","postrzępione",..: 1 1 1 2 1 1 1 2 1 2 ...
 $ I_Wyw            : Factor w/ 2 levels "brak","obecne": 2 2 2 2 2 2 2 2 2 2 ...
 $ I_SmolWyb        : Factor w/ 2 levels "brak","obecne": 2 2 2 1 2 1 2 1 2 2 ...
 $ N_Zabarwienie    : Factor w/ 5 levels "cały czerwonawy lub różowy",..: 5 4 5 5 1 1 5 1 5 1 ...

And the code for the sample pie chart is as follows (changed from: How to create a pie chart with percentage labels using ggplot2?):

> data <- mj %>% 
+     group_by(N_Zabarwienie) %>% 
+     count() %>% 
+     ungroup() %>% 
+     mutate(per=`n`/sum(`n`)) %>% 
+     arrange(desc(N_Zabarwienie))
> data$label <- scales::percent(data$per)
> ggplot(data=data)+
+     geom_bar(aes(x="", y=per, fill=N_Zabarwienie), stat="identity", width = 1)+
+     coord_polar("y", start=0)+
+     theme_void()+
+     geom_text(aes(x=1.3, y = cumsum(per) - per/2, label=label))

The chart looks like this:
Pie chart 1 - first code - all species

If I change the code as Benjamin Schlegel suggested:

> data <- mj %>% 
+     group_by(N_Zabarwienie) %>% 
+     count() %>% 
+     ungroup() %>% 
+     mutate(per=`n`/sum(`n`)) %>% 
+     arrange(desc(N_Zabarwienie))
> data$label <- scales::percent(data$per)
> ggplot(data=data)+
+     geom_bar(aes(x="", y=per, fill=N_Zabarwienie), stat="identity", width = 1)+
+     coord_polar("y", start=0)+
+     theme_void()+
+     geom_text(aes(x=1.3, y = cumsum(per) - per/2, label=label)) +
+     scale_fill_discrete(labels = c("zielony do bladego bez czerwieni", "zewnątrz blady", "w środku różowy", "cały czerwonawy lub różowy", "błyszcząco purpurowy"), drop = FALSE)

This chart looks like:
Pie chart 2 - second code - all species

In the first chart, the most common is "cały czerwonawy lub różowy", which means all reddish or pink (it is the color of the petiole), and in the second graph - "zielony do bladego bez czerwieni" which means green to pale without red. The difference is diametrical.

The first version is correct.

> summary(mj$N_Zabarwienie)
      cały czerwonawy lub różowy                na zewnątrz blady 
                            2943                                1 
           purpurowy z połyskiem                  w środku różowy 
                               1                              716 
zielony do bladego bez czerwieni 
                             732

As I change the scope of data (only for one of the species), it shows only part of the legend (the one that is currently in use).

Below is an example chart (first code) for the selected species (Taraxacum ancistrolobum).
Pie chart 3 - first code - Taraxacum ancistrolobum

And this is the same set of data, but generated from the second code.
Pie chart 4 - second code - Taraxacum ancistrolobum

And here also the first version is correct.

> summary(jta$N_Zabarwienie)
      cały czerwonawy lub różowy                na zewnątrz blady 
                             163                                0 
           purpurowy z połyskiem                  w środku różowy 
                               0                               85 
zielony do bladego bez czerwieni 
                             163

I would like to put charts made for different species next to each other and then compare them. A uniform legend is essential for it.

So I repeat the question:
how to make the same legend on all charts, despite different data ranges, but with default colors?

Do not use `scale_fill_manual`. – Roman Luštrik Jun 28 '18 at 09:20 — Roman Luštrik, Jun 28 '18 at 09:20
Thank you, but which command should I use? – Olga Świder Jun 28 '18 at 09:23 — Olga Świder, Jun 28 '18 at 09:23

score 2 · Accepted Answer · answered Jun 28 '18 at 09:42

2

scale_fill_manual, which is used in the question you liked to, is used, when the colors are getting defined. If you want the default colors, you can use scale_fill_discrete if your variable is discrete or scale_fill_continuous if your variable is numeric.

The example from the other question with specifying the colos would be:

ggplot(data =  dat, aes(x = Row, y = Col)) +
  geom_tile(aes(fill = Y1), color = "black") +
  scale_fill_discrete(labels = c("cat1", "cat2", "cat3", "cat4", "cat5", "cat6", "cat7", "cat8"), 
                      drop = FALSE)

answered Jun 28 '18 at 09:42

Benjamin Schlegel

525
4
10

You are welcome. Can you please upvote the question to show that it was useful for you? – Benjamin Schlegel Jun 28 '18 at 09:49
I am new here and I do not know what is going on. But if you explain: for the hero - everything ;-). – Olga Świder Jun 28 '18 at 09:52
Ah, I see, you need 15 reputation points to be able to upvote, but you only have 8. – Benjamin Schlegel Jun 28 '18 at 09:54
Thanks for the clarification. And for the answer. – Olga Świder Jun 28 '18 at 09:57
I looked closely at the charts and unfortunately there are errors in them. Therefore, I completed the question about my examples. – Olga Świder Jun 29 '18 at 09:53
You can leave the label parameter away, if you do not want to change the labels. If you want to changevthe label text, it must be in the order they are displayed in the legend. – Benjamin Schlegel Jun 30 '18 at 10:53
I checked it carefully this time. It works :-). Thank you very much (also for your patience ;-)). – Olga Świder Jul 02 '18 at 11:12

Default colors in ggplot2 for extended legend labels

1 Answers1