Geom_bar + geom_line with multiple variable in long format dataset

Question

Is there a way to have geom_line and geom_bar on the same plot without overlapping them?

I would like to have the variables "exp_j_perc" and "cut_j_perc" as bar while "total_growth" as line, but I can't figure out how to do it.

I tried this but I got the error that subset is an unknown parameter.

I would also have two different y axis, if possible.

df_long %>% 
    ggplot(aes(x = budget_year, y = value)) +
    geom_bar(subset = .(variable == "exp_j_perc" | variable == "cut_j_perc"),
             stat="identity", position = "dodge2", width = 0.35, alpha = 0.75)+
    geom_line(subset = .(variable == "total_growth"),
              size = 0.6, linetype = "longdash") +
    scale_x_continuous(breaks = c(2010,2011,2012,2013,2014,2015,2016,2017,2018)) +
    scale_fill_manual(values = palette) +
    xlab("Year") +
    theme_minimal() +
    theme(panel.grid.major.x = element_blank(),
          axis.text=element_text(size=10))

Here's the data:

df_long <- structure(list(year = c(2010, 2010, 2011, 2012, 2014, 2014, 2015, 
2016, 2018, 2010, 2010, 2011, 2012, 2014, 2014, 2015, 2016, 2018, 
2010, 2010, 2011, 2012, 2014, 2014, 2015, 2016, 2018, 2010, 2010, 
2011, 2012, 2014, 2014, 2015, 2016, 2018, 2010, 2010, 2011, 2012, 
2014, 2014, 2015, 2016, 2018, 2010, 2010, 2011, 2012, 2014), 
    budget_year = c(2010, 2011, 2012, 2013, 2014, 2015, 2016, 
    2017, 2018, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 
    2018, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 
    2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2010, 
    2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2010, 2011, 
    2012, 2013, 2014), variable = c("exp_j_perc", "exp_j_perc", 
    "exp_j_perc", "exp_j_perc", "exp_j_perc", "exp_j_perc", "exp_j_perc", 
    "exp_j_perc", "exp_j_perc", "cut_j_perc", "cut_j_perc", "cut_j_perc", 
    "cut_j_perc", "cut_j_perc", "cut_j_perc", "cut_j_perc", "cut_j_perc", 
    "cut_j_perc", "total_d", "total_d", "total_d", "total_d", 
    "total_d", "total_d", "total_d", "total_d", "total_d", "total_growth", 
    "total_growth", "total_growth", "total_growth", "total_growth", 
    "total_growth", "total_growth", "total_growth", "total_growth", 
    "exp_justification", "exp_justification", "exp_justification", 
    "exp_justification", "exp_justification", "exp_justification", 
    "exp_justification", "exp_justification", "exp_justification", 
    "cut_justification", "cut_justification", "cut_justification", 
    "cut_justification", "cut_justification"), value = c(33.33, 
    0, 22.45, 46.66, 50, 64, 82.35, 92.18, 100, 66.66, 100, 77.55, 
    53.33, 50, 36, 17.64, 7.82, 0, 1168782, 1136279.2186499, 
    1131940.80814086, 1151145.76656277, 1164509.2887094, 1181777.9454623, 
    1224866.3370881, 1265134.21792983, 1288684.9611902, NA, -2.78091049914321, 
    -0.381808488427549, 1.69663981400685, 1.16088878878757, 1.48291275306511, 
    3.64606496434037, 3.28753265743733, 1.86152132529547, 4, 
    0, 11, 21, 5, 16, 56, 59, 49, 8, 27, 38, 24, 5)), row.names = c(NA, 
-50L), class = c("tbl_df", "tbl", "data.frame"))

score 0 · Accepted Answer · answered Nov 18 '20 at 19:36

A quick scan of the documentation indicates that the subset argument doesn't exist in those specific functions and explains why that you're getting that error.

I suggest reading this post on what is possible, what is not suggested, and some opinions on data visualization. In the case of your data, it is possible for a two different scale y-axes because one of them represents a fixed value (%) so you can easily scale the other y-axis relative to it using the sec.axis argument in scales_y_continuous. Some pre-processing is required to separate the variables that you want to pass to each `geom_*

library(dplyr)
library(tidyr)
library(ggplot2)
df_long %>% 
  filter(variable %in% c("exp_j_perc", "cut_j_perc", "total_growth")) %>% 
  pivot_wider(names_from = variable) %>% 
  pivot_longer(cols = c("exp_j_perc", "cut_j_perc")) %>% 
  ggplot(aes(x = budget_year)) +
  geom_col(aes(y = value, fill = name)) +
  geom_line(aes(y = (total_growth*10) + 50), size = 2, alpha = 0.7, col = "blue") +
  scale_fill_discrete(name = "") +
  scale_y_continuous(sec.axis = sec_axis(~ (. -50)/10, name = "total growth")) +
  labs(y = "percent %")

I don't like this type of data viz but am acutely aware of some management having blinders on with is appropriate vs. not.

What if I want to add another line, and have the legend also for the geom_line? How can I do that? — user14514023, Nov 19 '20 at 13:04
that would require the similar data processing/transform steps taken with the variables being mapped in `geom_col()`, except now being fed into `geom_line` and adding an `aes(..., col = )` argument — EJJ, Nov 19 '20 at 15:53

Geom_bar + geom_line with multiple variable in long format dataset

1 Answers1