如何使用ggplot2制作条形图

Question

How to make a bar-plot like the picture below?

I've tried, but only can get this one:

Here is my code:

ggplot(N.Balance, aes(x = factor(Period), y = value)) + 
geom_bar(stat = "identity", aes( group = Type, fill = variable), osition = "stack", width = 0.6) +
facet_wrap(~ Type,ncol = 1) +
coord_flip() +
scale_fill_grey() + 
theme_bw(base_size = 30, base_family = "serif") + 
labs(y = expression(paste("kg", " ", "N", " ", ha^{-1}))) + 
theme(legend.key.height = unit(0.5, "in"))

And the data is shown below:

structure(list(Period = c("2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W", "2007R", "2007/2008W", "2008R", "2008/2009W", 
"2009R", "2009/2010W"), 
variable = c("Denitrification", "Denitrification", "Denitrification", "Denitrification",
"Denitrification", "Denitrification", "Runoff", "Runoff", "Runoff", "Runoff", "Runoff", 
"Runoff", "Leaching", "Leaching", "Leaching", "Leaching", "Leaching", "Leaching", "NH3Vol", 
"NH3Vol", "NH3Vol", "NH3Vol", "NH3Vol", "NH3Vol", "Harvest", 
"Harvest", "Harvest", "Harvest", "Harvest", "Harvest", "Fertilizer", 
"Fertilizer", "Fertilizer", "Fertilizer", "Fertilizer", "Fertilizer", 
"Fix", "Fix", "Fix", "Fix", "Fix", "Fix", "Irrigation", "Irrigation", 
"Irrigation", "Irrigation", "Irrigation", "Irrigation", "Seeds", 
"Seeds", "Seeds", "Seeds", "Seeds", "Seeds", "Deposition", "Deposition", 
"Deposition", "Deposition", "Deposition", "Deposition"), 
value = c(-89.4, -34.4, -61.5, -82.5, -87.2, -34.7, -21.8, -33.4, -2.65, -42.8, 
-19.2, -58.7, -8.22, -1.44, -9.76, -4.76, -4.97, -19, -71.6, 
-50.8, -97.1, -10.9, -60.6, -19.6, -187, -116, -167, -96, -177, 
-127, 300, 200, 300, 200, 300, 200, 45, 15, 45, 15, 45, 15, 12.5, 
0, 11.6, 0, 11.3, 0, 0.9, 3, 0.9, 3, 0.9, 3, 8.41, 13.74, 4.01, 
13.34, 16.31, 9.81), Type = c("O", "O", "O", "O", "O", "O", "O", 
"O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "O", 
"O", "O", "O", "O", "O", "O", "O", "O", "O", "O", "I", "I", "I", 
"I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I", 
"I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I", "I")), 
.Names = c("Period", "variable", "value", "Type"), class = "data.frame", 
row.names = c(NA,  -60L))

Note that the type "O" means "Output" and "I" means "Input"

Answer 1

It would have been more helpful to do the dput() I asked for in the comment, but I replaced the print() output with that for others who may want to contribute.

There are (at a minimum) two other ways (than the one below) to get close to that chart. This one relies on an emerging new geom_ that lets you have horizontal bars w/o coord_flip() - which means you can play with facet scales:

library(ggplot2)
library(ggstance) # devtools::install_github("lionel-/ggstance")
library(dplyr)

N.Balance <- read.csv("~/Data/so.csv", stringsAsFactors=FALSE)
N.Balance$Period_f <- factor(N.Balance$Period)
N.Balance$Type_f <- factor(N.Balance$Type, 
                         levels=c("O", "I"),
                         labels=c("Output", "Input"))

gg <- ggplot(N.Balance, aes(x=value, y=Period_f))
gg <- gg + geom_barh(stat = "identity", 
                    aes(group = Type, fill = variable), 
                    position = "stack", width = 0.6)
gg <- gg + geom_text(data=data.frame(Period=unique(N.Balance$Period)),
                     aes(x=5, y=Period, label=Period),
                     color="white", hjust=0, size=3)
gg <- gg + scale_x_continuous(expand=c(0,-0.001))
gg <- gg + scale_fill_grey(name="")
gg <- gg + facet_wrap(~ Type_f, ncol=2, scales="free_x")
gg <- gg + guides(fill=guide_legend(keywidth = 2, keyheight = 1))
gg <- gg + labs(y=NULL, x = expression(paste("kg", " ", "N", " ", ha^{-1})))
gg <- gg + theme_bw()
gg <- gg + theme(legend.key.height = unit(0.5, "in"))
gg <- gg + theme(panel.background=element_blank())
gg <- gg + theme(panel.border=element_blank())
gg <- gg + theme(panel.margin=margin(l=0, r=0))
gg <- gg + theme(legend.position="top")
gg <- gg + theme(panel.grid=element_blank())
gg <- gg + theme(strip.background=element_blank())
gg <- gg + theme(axis.text.y=element_blank())
gg <- gg + theme(axis.ticks.y=element_blank())
gg <- gg + theme(legend.key=element_blank())
gg <- gg + theme(legend.text=element_text(size=8))
gg <- gg + theme()
gg

This new geom_barh() is in a package that is not on CRAN (yet) so if that's an issue, there are—as I said—at least two other ways.

One other way gives you completely flush bars, and it's possible to do that w/just geom_bar() but it takes some data wrangling and axis text label wrangling. You have to hand craft what facet labeling gets you for free here (if you need the Input/Output labels).

If you need the legends separated (and some additional tweaking to get the bars flush), then most straighforward way is to make two separate plots, edit grob margins & use grid.arrange() . This method can also get you aligned Input/Output labels if you do it in the plot title.

Both of those are "work".

If you need pattern fills, someone else will have to help you. That's doable but tedious.