CODEWITHSUNDEEP
rggplot2
Simon Besnard
Simon Besnard

Reputation: 397

Plot two variables in the same histogram with ggplot

I have a dataframe and I want to make a histogram out of it by using ggplot. I want to have Forest, Agriculture, Built-up, Other and Water in the x-axis and in the y axis would be both comission and omission errors for each land use. Can someone help me out. Thanks 

 dput(n)
    structure(c(90.49964814, 36.77437804, 18.61958266, 57.14285714, 
    20, 100, 9.53346856, 45.63106796, 0, 0), .Dim = c(2L, 5L), .Dimnames = list(
        c("Omission Error", "Comission Error"), c("Forest", "Agriculture", 
        "Built-up", "Other", "Water")))

Upvotes: 2

Views: 8807

Answers (2)

KFB
KFB

Reputation: 3501

Or would this alternative fit?

library(reshape2); library(ggplot2)
df2 <- melt(df, id.var=Forest)
#                 X1          X2      value
# 1   Omission Error      Forest  90.499648
# 2  Comission Error      Forest  36.774378
# 3   Omission Error Agriculture  18.619583
# 4  Comission Error Agriculture  57.142857
# 5   Omission Error    Built-up  20.000000
# 6  Comission Error    Built-up 100.000000
# 7   Omission Error       Other   9.533469
# 8  Comission Error       Other  45.631068
# 9   Omission Error       Water   0.000000
# 10 Comission Error       Water   0.000000

ggplot(df2, aes(x=X2, y=value)) + geom_bar(stat="identity") + facet_grid(X1~.)

enter image description here

Upvotes: 4

Gregor Thomas
Gregor Thomas

Reputation: 145755

With ggplot, when you want things to share a dimension (e.g., x-axis, color), you want them in one column. So we start by reshaping your data into long form with reshape2::melt

library(reshape2)
library(scales) # for percent formatter
longdf <- melt(n)
names(longdf) <- c("Error", "Land_Use", "Measure")

Then the plotting is easy.

ggplot(longdf, aes(x = Land_Use, y = Measure / 100, fill = Error)) +
    geom_bar(stat = "identity", position = "dodge") +
    scale_y_continuous(labels = percent_format()) +
    labs(x = "Land Use", y = "Error Percent", fill = "Error Type")

Upvotes: 1

Related Questions