Here is the data frame:
> test
a b c
1 0.22904349 -0.12023869 0.1546898
2 1.09504754 -0.20398923 -0.9313251
3 -0.41200391 -0.16308791 0.6716105
4 -0.04356308 -1.81898245 -0.8074506
5 -1.23413459 1.24309479 -1.3861049
6 0.14266136 -2.22712577 -0.2341793
7 -0.25113445 0.60213281 -0.8106908
8 2.52372557 0.03794341 -1.4308955
9 0.66005867 0.74508029 -0.2922560
10 1.23552452 -0.26187445 -0.9874546
What I want to plot are the densities of a,b and c on a single graph. I want to be able to specify the columns to be plotted by their indices. Additionally, the densities can be colored according to their columns. This is the code I tried:
test<- as.data.frame(cbind(a=rnorm(1:10),b=rnorm(1:10),c=rnorm(1:10)))
for(i in seq(1,ncol(test),1)){
if(i==1)p<-ggplot(data=test, aes_string(x=names(test)[i]))
else p<-p+ggplot(data=test, aes_string(x=names(test)[i]))
}
p+geom_density()
Error I got:
Error in p + o : non-numeric argument to binary operator
In addition: Warning message:
Incompatible methods ("+.gg", "Ops.data.frame") for "+"
Please advice. Thanks
The standard ggplot way is to use long data, not wide data:
library(tidyr)
test_long = gather(test)
ggplot(test_long, aes(x = value, color = key)) +
geom_density()
If you really want indices in there, we'll add them to the long data:
test_long$index = match(test_long$key, names(test))
And then the way to select which ones to plot is to subset the data passed to ggplot
# if you only want columns 2 and 3 from the original data
ggplot(test_long[test_long$index %in% c(2, 3), ],
aes(x = value, color = key)) +
geom_density()
And, if you really want to be stubborn, the problem with your for loop is that ggplot is called multiple times. ggplot() initializes a plot, you can't add it to a plot multiple times. You can fix it, but you shouldn't do things this way.
p = ggplot(data = test)
for(i in seq_along(test)) {
if (i == 1) p = p + geom_density(aes_string(x = names(test)[i]))
else p = p + geom_density(aes_string(x = names(test)[i]), color = "green")
}
print(p)
In this case ggplot isn't being used as intended so you'd have to set up your own colors and adding a legend will be a real pain. Which is part of why you should do it the other way, the easy way.
Edits: In a fresh R session, this runs just fine for me:
# load packages
library(tidyr)
library(ggplot2)
# data from the question
test <- as.data.frame(cbind(a=rnorm(1:10),b=rnorm(1:10),c=rnorm(1:10)))
# long format
test_long = gather(test)
# plot all 3
ggplot(test_long, aes(x = value, color = key)) +
geom_density()
# add original data indices
test_long$index = match(test_long$key, names(test))
# plot only columns 2 and 3
ggplot(test_long[test_long$index %in% c(2, 3), ],
aes(x = value, color = key)) +
geom_density()