Wrong usage of R package reshape2 to plot with ggplot2?

Question

I would like to plot some data with ggplot2. My data is originally in wide format, and so I would like to transform it to long format first.

Somehow "melt" produces a table I cannot use with ggplot2 as desired. I guess I am using it wrong, but I could sadly not find out, how to solve the problem.

I would like to plot the variables s1 and s2 vs the wavelength. This is a minimal half working example:

#Generate fake data:
data <- cbind(wavelength = seq(1,10), s1 = seq(21,30), s3 = seq(41,50))

#Convert to long format:
#install.packages("reshape2")
library(reshape2)
datLong <- melt(data          = data,
            id.vars       = c("wavelength"),
            measure.vars  = c("s1","s2"),
            variable.name = "variable",
            value.name    = "value")

#install.packages("ggplot2")
library("ggplot2")

#The following delivers the error "Error: Attempted to create layer with no stat."
 ggplot(data = datLong, mapping = aes(x = wavelength, y = value)) + layer(geom = "point") 

#This plots something, but it seems that "melt" produced a wrong format as wavelength, s1, s2 are on the x-axis.
ggplot() + layer(data = datLong, mapping = aes(x=Var2, y=value), geom = "point", stat = "identity", position_dodge(width = 3))

It would be very great if you could explain the problem.

Thank you very much!

The issue is that data is a matrix. So you are actually using melt.array and not melt.data.frame. The argument names differ and hence the result is not what you are looking for, see ?reshape2::melt for details. — markus
– markus, Commented Jan 25, 2019 at 8:50
I'm not sure but it's rare to find a layer created as in case 1. Normally you just call the geom as done in my answer. Two fails because you set the x-axis to Var2(which isn't wavelength I think). — NelsonGon
– NelsonGon, Commented Jan 25, 2019 at 8:52

NelsonGon · Accepted Answer · 2019-01-25 08:49:19Z

1

Is this what you had in mind?

library(ggplot2)
library(dplyr)
library(tidyr)
#or just library(tidyverse)
as.tibble(data) %>% 
  gather("Var","Val",-wavelength) %>% 
  ggplot(mapping = aes(x = wavelength, y = Val,col=Var)) + geom_point()

Using reshape2 and the tidyverse

as.tibble(data) %>% 
  melt(id.vars=c("wavelength")) %>% 
  ggplot(mapping = aes(x = wavelength, y = value,col=variable)) + geom_point()

edited Jan 25, 2019 at 8:49

answered Jan 25, 2019 at 8:44

NelsonGon

13.3k7 gold badges32 silver badges60 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

Paul Over a year ago

exactly what I was going to do, except maybe use as_tibble rather than as.data.frame to keep it in the tidyverse family.

NelsonGon Over a year ago

Edited. Thanks! Used as.tibble though.

Paul Over a year ago

I'm just being picky now, but as.tibble is depreciated.

Roland Over a year ago

Thank you very much! Works perfect! Still I am interested in what was the mistake I made.

NelsonGon Over a year ago

@Roland it was in a non data.frame format. I've made another comment below the question. You can see this [,1] in the original code you provided. This suggests the data is not a dataframe. You can use str to see the object's data type. Use class(data) to see its class

|

Collectives™ on Stack Overflow

Wrong usage of R package reshape2 to plot with ggplot2?

1 Answer 1

6 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

6 Comments

Your Answer

Sign up or log in

Post as a guest

Related