I'm trying to create a horizontal boxplot with logarithmic axis using ggplot2. But, the length of whiskers are wrong.
A minimal reproducible example:
Some data
library(ggplot2)
library(reshape2)
set.seed(1234)
my.df <- data.frame(a = rnorm(1000,150,50), b = rnorm(1000,500,150))
my.df$a[which(my.df$a < 5)] <- 5
my.df$b[which(my.df$b < 5)] <- 5
If I plot this using base R boxplot(), everything is fine
boxplot(my.df, log="x", horizontal=T)
But with ggplot,
my.df.long <- melt(my.df, value.name = "vals")
ggplot(my.df.long, aes(x=variable, y=vals)) +
geom_boxplot() +
scale_y_log10(breaks=c(5,10,20,50,100,200,500,1000), limits=c(5,1000)) +
theme_bw() + coord_flip()
I get this plot, in which the whiskers are the wrong length (see for example how there are many additional outliers below the whiskers and none above).
Note that, without log axes, ggplot has the whiskers the correct length
ggplot(my.df.long, aes(x=variable, y=vals)) +
geom_boxplot() +
theme_bw() + coord_flip()
How do I produce a horizontal logarithmic boxplot using ggplot with the correct length whiskers? Preferably with the whiskers extending to 1.5 times the IQR.
N.B. as explained here. It is possible to use coord_trans(y = "log10") instead of scale_y_log10, which will cause the stats to be calculated before transforming the data. However, coord_trans cannot be used in combination with coord_flip. So this does not solve the issue of creating horizontal boxplots with a log axis.
You can have ggplot use boxplot.stats (the same function used by base boxplot) to set the y-values for the box-and-whiskers and the outliers. For example:
# Function to use boxplot.stats to set the box-and-whisker locations
mybxp = function(x) {
bxp = boxplot.stats(x)[["stats"]]
names(bxp) = c("ymin","lower", "middle","upper","ymax")
return(bxp)
}
# Function to use boxplot.stats for the outliers
myout = function(x) {
data.frame(y=boxplot.stats(x)[["out"]])
}
Now we use those functions in stat_summary to draw the boxplot, as in the example below:
ggplot(my.df.long, aes(x=variable, y=vals)) +
stat_summary(fun.data=mybxp, geom="boxplot") +
stat_summary(fun.data=myout, geom="point") +
theme_bw() + coord_flip()
Now for the log transformation issue: The plots below show, respectively, no coordinate transformation, scale_y_log10, and coord_trans(y="log10"). In addition, I've used geom_hline to add dotted lines at each of the box-and-whisker values and I've added text to show the actual values. To reduce clutter, I've removed the outlier points, and I've faded out the boxplots a bit so that the other components will show up better.
# Set up common plot elements
p = ggplot(my.df.long, aes(x=variable, y=vals)) +
geom_hline(yintercept=mybxp(my.df$a), colour="red", lty="11", size=0.3) +
geom_hline(yintercept=mybxp(my.df$b), colour="blue", lty="11", size=0.3) +
stat_summary(fun.data=mybxp, geom="boxplot", colour="#000000A0", fatten=0.5) +
#stat_summary(fun.data=myout, geom="point") +
theme_bw() + coord_flip()
br = c(5,10,20,50,100,200,500,1000)
## Create plots
# Without log transformation
p1 = p + scale_y_continuous(breaks=br, limits=c(5,1000)) +
stat_summary(fun.y=mybxp, aes(label=round(..y..)), geom="text", size=3, colour="red") +
ggtitle("No Transformation")
# With scale_y_log10
p2 = p + scale_y_log10(breaks=br, limits=c(5,1000)) + ggtitle("scale_y_log10") +
stat_summary(fun.y=mybxp, aes(label=round(..y..,2)), geom="text", size=3, colour="red") +
stat_summary(fun.y=mybxp, aes(label=round(10^(..y..))), geom="text", size=3,
colour="blue", position=position_nudge(x=0.3))
# With coord_trans
p3 = p + scale_y_continuous(breaks=br, limits=c(5,1000)) +
stat_summary(fun.y=mybxp, aes(label=round(..y..)), geom="text", size=3, colour="red") +
coord_trans(y="log10") + ggtitle("coord_trans(y='log 10')")
The three plots are shown below. Note that the last plot, using coord_trans is not flipped, because coord_trans overrides coord_flip. You can probably use something like the code in this SO answer to flip the plot, but I haven't done that here.
The first plot, with no transformations, shows the correct values.
The third plot, using coord_trans also has everything in the correct locations. Note that coord_trans is actually changing the y-coordinate system of the plot without changing the values of the plotted points. It's the space itself that's been "distorted" to a log scale.
Now, note that in the second plot, using scale_y_log10, the boxes are in the correct locations but the ends of the whiskers are in the wrong locations. On the other hand, comparison with the other two plots shows that the location of all the geom_hlines is correct. Also note that, unlike coord_trans, scale_y_log10 takes the log of the points themselves and just relabels the y-axis breaks with the unlogged values, while leaving the "space" in the which the points are plotted unchanged. You can see this by looking at the values in red text. The values in blue text are the unlogged values.
See #dww's answer for an explanation of why scale_y_log10 results only in the whisker ends being transformed incorrectly, while the box values are plotted in the right place.
The problem is due to the fact that scale_y_log10 transforms the data before calculating the stats. This does not matter for the median and percentile points, because e.g. 10^log10(median) is still the median value, which will be plotted in the correct location. But it does matter for the whiskers which are calculated using 1.5 * IQR, because 10^(1.5 * IQR(log10(x)) is not equal to 1.5 * IQR(x). So the calculation fails for the whiskers.
This error becomes evident if we compare
boxplot.stats(my.df$b)$stats
# [1] 117.4978 407.3983 502.0460 601.2937 873.0992
10^boxplot.stats(log10(my.df$b))$stats
# [1] 231.1603 407.3983 502.0459 601.2935 975.1906
In which we see that the median and percentile ppoints are identical, but the whisker ends (1st and last elements of the stats vector) differ
This detailed and useful answer by #eipi10, shows how to calculate the stats yourself and force ggplot to use these user-defined stats rather than its internal (and incorrect) algorithm. Using this approach, it becomes relatively simple to calculate the correct statistics and use these instead.
# Function to use boxplot.stats to set the box-and-whisker locations
mybxp = function(x) {
bxp = log10(boxplot.stats(10^x)[["stats"]])
names(bxp) = c("ymin","lower", "middle","upper","ymax")
return(bxp)
}
# Function to use boxplot.stats for the outliers
myout = function(x) {
data.frame(y=log10(boxplot.stats(10^x)[["out"]]))
}
ggplot(my.df.long, aes(x=variable, y=vals)) + theme_bw() + coord_flip() +
scale_y_log10(breaks=c(5,10,20,50,100,200,500,1000), limits=c(5,1000)) +
stat_summary(fun.data=mybxp, geom="boxplot") +
stat_summary(fun.data=myout, geom="point")
Which produces the correct plot
A note on using coord_trans as an alternative approach:
Using coord_trans(y = "log10") instead of scale_y_log10, causes the stats to be calculated (correctly) on the untransformed data. However, coord_trans cannot be used in combination with coord_flip. So, this does not solve the issue of creating horizontal boxplots with a log axis. The suggestion here to use ggdraw(switch_axis_position()) from the cowplot package to flip the axes after using coord_trans did not work, but throws an error (cowplot v0.4.0 with ggplot2 v2.1.0)
Error in Ops.unit(gyl$x, grid::unit(0.5, "npc")) : both operands
must be units
In addition: Warning message: axis.ticks.margin is
deprecated. Please set margin property of axis.text instead
I think that the easiest answer if you don't need to make the boxplots horizontal is to transform the coordinate system in stead of changing the scale, using coord_trans(y = "log10") in stead of scale_y_log10().
Related
Setting the ggplot2 binwidth in geom_histogram while using scale_x_log10 produces a weird histogram.
I want to adjust the binwidth without the workaround found here.
One reason I don't want to use the workaround is that I don't like it; it seems like there ought to be a better way built into ggplot. The other reason is that it didn't work when I tried it on my data set.
I'm using facet_wrap, so the solution needs to work with that, but the example code I'm using is stripped down to the minimum.
When I allow the default binwidth, I get a decent histogram:
library(ggplot2)
data(diamonds)
ggplot(data=diamonds, aes(x=price/carat)) +
geom_histogram() +
scale_x_log10()# +
# facet_wrap(~cut, ncol=1, scales='free_y')
But, when I set the binwidth, I get a uniform distribution filling the entire graph (or a single bin?) no matter what the binwidth (except when binwidth=1, which produces what look like two bins, or a bimodal uniform distribution?):
ggplot(data=diamonds, aes(x=price/carat)) +
geom_histogram(binwidth=10) +
scale_x_log10()# +
# facet_wrap(~cut, ncol=1, scales='free_y')
Setting breaks produces the same filled square with new breaks. Setting limits the clears the graph.
Setting the binwidth from within ggplot() itself leaves the graph unchanged from default binwidths, presumably because geom_histogram overrides it. And, scale_x_log10 doesn't accept binwidth.
It works to set binwidth while using scale_x_continuous instead of scale_x_log10.
Try entering a fraction of the total width, such that the binwidth relates to number of bins as something like 1/(n_bins - 1).
library(ggplot2)
data(diamonds)
ggplot(data=diamonds, aes(x=price/carat)) +
geom_histogram(binwidth = 1/50) +
scale_x_log10()
Is there a workaround for when one wants to apply geom_rect to Infinity on the y axis of a ggplot object when a transformation is applied to the y axis?
The code below does not plot the interval rectangles unless you comment out the scale_y_continuous line. When using the transformed scale, I have to put in actual data limits. I could probably write a function to find the min/max of the other data being plotted to avoid hard coding values but I'm looking for something closer to the Inf approach. I tried using NA instead of Inf but no luck.
library(tidyverse)
data(economics)
ints<-data.frame(start=as.Date(paste0(seq(1970,2020,by=10),"-01-01"))) %>%
mutate(end=start+1785)
plt<-ggplot(economics,aes(date,unemploy)) + theme_bw() +
scale_y_continuous(trans="sqrt") +
geom_rect(data=ints,inherit.aes=F,aes(xmin=start,xmax=end,ymin=-Inf,ymax=Inf)) + geom_line()
plt
library(tidyverse)
data(economics)
ints<-data.frame(start=as.Date(paste0(seq(1970,2020,by=10),"-01-01"))) %>%
mutate(end=start+1785)
plt<-ggplot(economics,aes(date,unemploy)) + theme_bw() +
scale_y_continuous(trans="sqrt") +
geom_rect(data=ints,inherit.aes=F,aes(xmin=start,xmax=end,ymin=0,ymax=Inf)) + geom_line()+
coord_cartesian(ylim=c(2000,20000)) # This will allow you to control how zoomed in you want the plot
plt
I'm plotting seasonal data using ggplot2. Since small values aren't visible well when I start the axis at 0 I want to offest the axis. I want the function values correspond to the area between the offset and the line, the transform for this would be r = sqrt(y+offest^2). (Correct me if I'm wrong here.)
How can I use this transformation together with geom_polar? In the following Code the Transformation does not seem to be applied:
my_function <- function(x){1+sin(x)}
my_trafo_trans <- function() trans_new("my_trafo", function(y){sign(y)*sqrt(abs(y))+1^2}, function(r){r^2+1^2})
ggplot(data=data.frame(x=0), aes(x=0)) +
coord_trans(y = "my_trafo") +
coord_polar(theta="x") +
ylim(-0.5, 2 ) +
xlim(-pi, pi) +
stat_function(fun=my_function, geom="line", alpha=0.75, n=1000) +
stat_function(fun=function(x) 0, geom="line") +
theme_bw()
Compare the output of this with the output when you comment out the coord_trans and coord_polar lines respectively.
Is it even possible to use coord_trans and coord_polar together?
The following code
library(ggplot2)
library(reshape2)
m=melt(iris[,1:4])
ggplot(m, aes(value)) +
facet_wrap(~variable,ncol=2,scales="free_x") +
geom_histogram()
produces 4 graphs with fixed y axis (which is what I want). However, by default, the y axis is only displayed on the left side of the faceted graph (i.e. on the side of 1st and 3rd graph).
What do I do to make the y axis show itself on all 4 graphs? Thanks!
EDIT: As suggested by #Roland, one could set scales="free" and use ylim(c(0,30)), but I would prefer not to have to set the limits everytime manually.
#Roland also suggested to use hist and ddply outside of ggplot to get the maximum count. Isn't there any ggplot2 based solution?
EDIT: There is a very elegant solution from #babptiste. However, when changing binwidth, it starts to behave oddly (at least for me). Check this example with default binwidth (range/30). The values on the y axis are between 0 and 30,000.
library(ggplot2)
library(reshape2)
m=melt(data=diamonds[,c("x","y","z")])
ggplot(m,aes(x=value)) +
facet_wrap(~variable,ncol=2,scales="free") +
geom_histogram() +
geom_blank(aes(y=max(..count..)), stat="bin")
And now this one.
ggplot(m,aes(x=value)) +
facet_wrap(~variable,scales="free") +
geom_histogram(binwidth=0.5) +
geom_blank(aes(y=max(..count..)), stat="bin")
The binwidth is now set to 0.5 so the highest frequency should change (decrease in fact, as in tighter bins there will be less observations). However, nothing happened with the y axis, it still covers the same amount of values, creating a huge empty space in each graph.
[The problem is solved... see #baptiste's edited answer.]
Is this what you're after?
ggplot(m, aes(value)) +
facet_wrap(~variable,scales="free") +
geom_histogram(binwidth=0.5) +
geom_blank(aes(y=max(..count..)), stat="bin", binwidth=0.5)
ggplot(m, aes(value)) +
facet_wrap(~variable,scales="free") +
ylim(c(0,30)) +
geom_histogram()
Didzis Elferts in https://stackoverflow.com/a/14584567/2416535 suggested using ggplot_build() to get the values of the bins used in geom_histogram (ggplot_build() provides data used by ggplot2 to plot the graph). Once you have your graph stored in an object, you can find the values for all the bins in the column count:
library(ggplot2)
library(reshape2)
m=melt(iris[,1:4])
plot = ggplot(m) +
facet_wrap(~variable,scales="free") +
geom_histogram(aes(x=value))
ggplot_build(plot)$data[[1]]$count
Therefore, I tried to replace the max y limit by this:
max(ggplot_build(plot)$data[[1]]$count)
and managed to get a working example:
m=melt(data=diamonds[,c("x","y","z")])
bin=0.5 # you can use this to try out different bin widths to see the results
plot=
ggplot(m) +
facet_wrap(~variable,scales="free") +
geom_histogram(aes(x=value),binwidth=bin)
ggplot(m) +
facet_wrap(~variable,ncol=2,scales="free") +
geom_histogram(aes(x=value),binwidth=bin) +
ylim(c(0,max(ggplot_build(plot)$data[[1]]$count)))
It does the job, albeit clumsily. It would be nice if someone improved upon that to eliminate the need to create 2 graphs, or rather the same graph twice.
Summary: I want to choose the colors for a ggplot2() density distribution plot without losing the automatically generated legend.
Details: I have a dataframe created with the following code (I realize it is not elegant but I am only learning R):
cands<-scan("human.i.cands.degnums")
non<-scan("human.i.non.degnums")
df<-data.frame(grp=factor(c(rep("1. Candidates", each=length(cands)),
rep("2. NonCands",each=length(non)))), val=c(cands,non))
I then plot their density distribution like so:
library(ggplot2)
ggplot(df, aes(x=val,color=grp)) + geom_density()
This produces the following output:
I would like to choose the colors the lines appear in and cannot for the life of me figure out how. I have read various other posts on the site but to no avail. The most relevant are:
Changing color of density plots in ggplot2
Overlapped density plots in ggplot2
After searching around for a while I have tried:
## This one gives an error
ggplot(df, aes(x=val,colour=c("red","blue"))) + geom_density()
Error: Aesthetics must either be length one, or the same length as the dataProblems:c("red", "blue")
## This one produces a single, black line
ggplot(df, aes(x=val),colour=c("red","green")) + geom_density()
The best I've come up with is this:
ggplot() + geom_density(aes(x=cands),colour="blue") + geom_density(aes(x=non),colour="red")
As you can see in the image above, that last command correctly changes the colors of the lines but it removes the legend. I like ggplot2's legend system. It is nice and simple, I don't want to have to fiddle about with recreating something that ggplot is clearly capable of doing. On top of which, the syntax is very very ugly. My actual data frame consists of 7 different groups of data. I cannot believe that writing + geom_density(aes(x=FOO),colour="BAR") 7 times is the most elegant way of coding this.
So, if all else fails I will accept with an answer that tells me how to get the legend back on to the 2nd plot. However, if someone can tell me how to do it properly I will be very happy.
set.seed(45)
df <- data.frame(x=c(rnorm(100), rnorm(100, mean=2, sd=2)), grp=rep(1:2, each=100))
ggplot(data = df, aes(x=x, color=factor(grp))) + geom_density() +
scale_color_brewer(palette = "Set1")
ggplot(data = df, aes(x=x, color=factor(grp))) + geom_density() +
scale_color_brewer(palette = "Set3")
gives me same plots with different sets of colors.
Provide vector containing colours for the "values" argument to map discrete values to manually chosen visual ones:
ggplot(df, aes(x=val,color=grp)) +
geom_density() +
scale_color_manual(values=c("red", "blue"))
To choose any colour you wish, enter the hex code for it instead:
ggplot(df, aes(x=val,color=grp)) +
geom_density() +
scale_color_manual(values=c("#f5d142", "#2bd63f")) # yellow/green