ggplot: Subset a layer where data is passed using a pipe

Question

I am trying to subset a layer of a plot where I am passing the data to ggplot through a pipe.

Here is an example:

library(dplyr)
library(ggplot2)
library(scales)

set.seed(12345)
df_example = data_frame(Month = rep(seq.Date(as.Date("2015-01-01"),
                                             as.Date("2015-12-31"), by = "month"), 2),
                        Value = sample(seq.int(30, 150), size = 24, replace = TRUE),
                        Indicator = as.factor(rep(c(1, 2), each = 12)))

df_example %>% 
  group_by(Month) %>% 
  mutate(`Relative Value` = Value/sum(Value)) %>% 
  ungroup() %>% 
  ggplot(aes(x = Month, y = Value, fill = Indicator, group = Indicator)) + 
  geom_bar(position = "fill", stat = "identity") + 
  theme_bw()+ 
  scale_y_continuous(labels = percent_format()) + 
  geom_line(aes(x = Month, y = `Relative Value`))

This gives:

I would like only one of those lines to appear, which I would be able to do if something like this worked in the geom_line layer:

  geom_line(subset = .(Indicator == 1), aes(x = Month, y = `Relative Value`))

Edit:

Session info:

R version 3.2.1 (2015-06-18) Platform: x86_64-w64-mingw32/x64 (64-bit) Running under: Windows Server 2012 x64 (build 9200)

locale: 2 LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252 [3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C [5] LC_TIME=English_United States.1252

attached base packages: 2 stats graphics grDevices utils
datasets methods base

other attached packages: 2 scales_0.3.0 lubridate_1.3.3 ggplot2_1.0.1 lazyeval_0.1.10 dplyr_0.4.3 RSQLite_1.0.0
readr_0.2.2 [8] RJDBC_0.2-5 DBI_0.3.1 rJava_0.9-7

loaded via a namespace (and not attached): 2 Rcpp_0.12.2
knitr_1.11 magrittr_1.5 MASS_7.3-40 munsell_0.4.2
lattice_0.20-31 [7] colorspace_1.2-6 R6_2.1.1 stringr_1.0.0 plyr_1.8.3 tools_3.2.1 parallel_3.2.1 [13] grid_3.2.1
gtable_0.1.2 htmltools_0.2.6 yaml_2.1.13 assertthat_0.1
digest_0.6.8 [19] reshape2_1.4.1 memoise_0.2.1
rmarkdown_0.8.1 labeling_0.3 stringi_1.0-1 zoo_1.7-12
[25] proto_0.3-10

I don't get the same plot as you, my lines are scaled quite differently. Also you should set a random seed so we can all work with the same plot. — Mike Wise
– Mike Wise, Commented Dec 26, 2015 at 8:42
Ok, reinitialized my workspace and the scale issue went away. Was some wierd side effect of earlier ggplot calls. — Mike Wise
– Mike Wise, Commented Dec 26, 2015 at 8:49
@MikeWise Yeah, I figured. The piped data should clearly be available down the line to be used with subset, but the usual suspects such as . do not appear to work. @Hadley Halp? — tchakravarty
– tchakravarty, Commented Dec 26, 2015 at 8:52

mgiormenti · Accepted Answer · 2018-11-15 18:29:12Z

tl;dr: Pass the data to that layer as a function that subsets the plot's data according to your criteria.

According to ggplots documentation on layers, you have 3 options when passing the data to a new layer:

If NULL, the default, the data is inherited from the plot data as specified in the call to ggplot().

A data.frame, or other object, will override the plot data. All objects will be fortified to produce a data frame. See fortify() for which variables will be created.

A function will be called with a single argument, the plot data. The return value must be a data.frame, and will be used as the layer data.

The first two options are the most usual ones, but the 3rd is perfect for our needs when the data has been modified through pyps.

In your example, adding data = function(x) subset(x,Indicator == 1) to the geom_line does the trick:

library(dplyr)
library(ggplot2)
library(scales)

set.seed(12345)
df_example = data_frame(Month = rep(seq.Date(as.Date("2015-01-01"),
                                             as.Date("2015-12-31"), by = "month"), 2),
                        Value = sample(seq.int(30, 150), size = 24, replace = TRUE),
                        Indicator = as.factor(rep(c(1, 2), each = 12)))

df_example %>% 
  group_by(Month) %>% 
  mutate(`Relative Value` = Value/sum(Value)) %>% 
  ungroup() %>% 
  ggplot(aes(x = Month, y = Value, fill = Indicator, group = Indicator)) + 
  geom_bar(position = "fill", stat = "identity") + 
  theme_bw()+ 
  scale_y_continuous(labels = percent_format()) + 
  geom_line(data = function(x) subset(x,Indicator == 1), aes(x = Month, y = `Relative Value`))

This is the resulting plot

Mike Wise · Accepted Answer · 2015-12-26 11:39:06Z

2

library(dplyr)
library(ggplot2)
library(scales)

set.seed(12345)
df_example = data_frame(Month = rep(seq.Date(as.Date("2015-01-01"),
                                             as.Date("2015-12-31"), by = "month"), 2),
                        Value = sample(seq.int(30, 150), size = 24, replace = TRUE),
                        Indicator = as.factor(rep(c(1, 2), each = 12)))

df_example %>% 
  group_by(Month) %>% 
  mutate(`Relative Value` = Value/sum(Value)) %>% 
  ungroup() %>% 
  ggplot(aes(x = Month, y = Value, fill = Indicator, group = Indicator)) + 
  geom_bar(position = "fill", stat = "identity") + 
  theme_bw()+ 
  scale_y_continuous(labels = percent_format()) + 
  geom_line(aes(x = Month, y = `Relative Value`,linetype=Indicator)) +
  scale_linetype_manual(values=c("1"="solid","2"="blank"))

yields:

edited Dec 26, 2015 at 11:39

answered Dec 26, 2015 at 8:56

Mike Wise

22.9k10 gold badges85 silver badges107 bronze badges

5 Comments

tchakravarty Over a year ago

Haha, just hide one of the lines. I like it. Let me see if there is another way to do this, else will mark this. Still curious about how to access piped data (basically not explicitly named data) within ggplot layers.

Mike Wise Over a year ago

Pipes don't give you much chance to hack the data on the fly.

Mike Wise Over a year ago

Down votes don't give much feedback - especially 18 months after the fact. How about a comment instead so I have a chance to fix whatever is bothering you?

kmace Over a year ago

This answer doesn't answer the spirit of the question. It works for his particular case, but not in general

Mike Wise Over a year ago

Wtf does "spirit of the question" mean anyway? Seems like the OP would know better than you. Actually it looks to me like a change in something (either ggplot or R) has broken my answer - it simply does not work now like it used to. I might get around to fixing it on the weekend. If you had actually tried my code you would have seen that.

Elio Campitelli · Accepted Answer · 2018-04-28 01:35:01Z

You might benefit from stat_subset(), a stat I made for my personal use that is available in metR: https://eliocamp.github.io/metR/articles/Visualization-tools.html#stat_subset

It has an aesthetic called subset that takes a logical expression and subsets the data accordingly.

library(dplyr)
library(ggplot2)
library(scales)

set.seed(12345)
df_example = data_frame(Month = rep(seq.Date(as.Date("2015-01-01"),
                                             as.Date("2015-12-31"), by = "month"), 2),
                        Value = sample(seq.int(30, 150), size = 24, replace = TRUE),
                        Indicator = as.factor(rep(c(1, 2), each = 12)))

df_example %>% 
   group_by(Month) %>% 
   mutate(`Relative Value` = Value/sum(Value)) %>% 
   ungroup() %>% 
   ggplot(aes(x = Month, y = Value, fill = Indicator, group = Indicator)) + 
   geom_bar(position = "fill", stat = "identity") + 
   theme_bw()+ 
   scale_y_continuous(labels = percent_format()) + 
   metR::stat_subset(aes(x = Month, y = `Relative Value`, subset = Indicator == 1), 
               geom = "line")

Collectives™ on Stack Overflow

ggplot: Subset a layer where data is passed using a pipe

Edit:

3 Answers 3

1 Comment

5 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

Edit:

3 Answers 3

1 Comment

5 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related