Barplot using ggplot

Question

I have a data frame like

  Model1    Model2      Model3    Model4
      4          4          5         5
      4          4         NA        NA
      3          2          5         5
      2          2          3         3
      3          3          3         3
      3          3          4         4

I want to plot a barplot like

enter image description here

The x- axis will have the Model1,Model2,Model3 and Model4 and the bars will be the proportion of the scores in each column,i.e., 6 bars for each Model (in the full data I have the scores from 0 to 5)

How can this be done without having to create a dataframe using rbind or creating a matrix? Any help will be appreciated.

Have you tried anything yet? Have a look at this post stackoverflow.com/questions/22189435/… and adapt it accordingly — infominer
– infominer, Commented Mar 5, 2014 at 18:04
@infominer I am not so proficient in R. if you could shed some light given my problem — darkage
– darkage, Commented Mar 5, 2014 at 18:10
Why do you not want to transform your data? It would then be straight forward to plot using ggplot. — user20650
– user20650, Commented Mar 5, 2014 at 18:23

Christie Haskell Marsh · Accepted Answer · 2014-03-05 20:17:39Z

2

You can't do it without changing the format of your data, but it isn't difficult to do so.

You can use the melt option of the reshape2 package.

I created a data frame of your data:

   df1<-data.frame("Model1"=c(4,4,3,2,3,3),
                   "Model2"=c(4,4,2,2,3,3),
                   "Model3"=c(5,NA,5,3,3,4),
                   "Model4"=c(5,NA,5,3,3,4))

  Model1 Model2 Model3 Model4
1      4      4      5      5
2      4      4     NA     NA
3      3      2      5      5
4      2      2      3      3
5      3      3      3      3
6      3      3      4      4

Next you can reshape the data using the reshape2 package:

library(reshape2)
df2<-melt(df1)

    variable value
1     Model1     4
2     Model1     4
3     Model1     3
4     Model1     2
5     Model1     3
6     Model1     3
7     Model2     4
8     Model2     4
9     Model2     2
10    Model2     2
11    Model2     3
12    Model2     3
13    Model3     5
14    Model3    NA
15    Model3     5
16    Model3     3
17    Model3     3
18    Model3     4
19    Model4     5
20    Model4    NA
21    Model4     5
22    Model4     3
23    Model4     3
24    Model4     4

Renamed the columns for ease:

names(df2)<-c("Model","Score")

Then calculate the proportions:

df3 <- as.data.frame(table(df2))
df3$prop<-df4$Freq/4*100

And finally, the plot:

ggplot(df3,aes(x=Model,y=prop,fill=as.factor(Score)))+
       geom_bar(stat="identity",position="dodge")+
       xlab("Models")+
       ylab("Prop. of Cases (%)")+
       ggtitle("Sample Data")+
       guides(fill=guide_legend(title="Scores"))+
       scale_fill_manual(values = c("2" = "lightblue", "3" = "mistyrose","4"="lightcyan","5"="lavender"))+
       theme_bw()

edited Mar 5, 2014 at 20:17

answered Mar 5, 2014 at 18:41

Christie Haskell Marsh

2,23420 silver badges18 bronze badges

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

6 Comments

darkage Over a year ago

I want the scores , i.e., 0,1,2,3,4,5 in the legend and the proportion in the y-axis

Christie Haskell Marsh Over a year ago

Edited my answer. Like that?

darkage Over a year ago

yes correct, but the y-axis should have the proportions,say for example in model1 2 occurs in 20% cases and similarly for other categories. One more question, will it be a good idea to have the row variable created in case I have 500K records?

Christie Haskell Marsh Over a year ago

The row variable isn't necessary unless you want to at some point refer to each value for each model based on its row position. If the order of the values under each model isn't relevant, then a row name is also likely to be irrelevant. Stand by on the proportions.

darkage Over a year ago

I tried the following: 1. created a seperate dataframe having two columns, viz. Score and Model 2. created a matix with the proportions mat <- table(Q$Score,Q$Model) mat <- round(prop.table(mat,margin=2)*100,2) 3. used barplot to plot : barplot(mat, beside = TRUE, ` col = c("lightblue", "mistyrose", "lightcyan", "lavender",` "cornsilk","green"), ` legend.text = rownames(mat), ylim = c(0, 40),main="Sample` Data", ` xlab="Models",ylab="Prop. of Cases (%)")` It worked

|

Collectives™ on Stack Overflow

Barplot using ggplot

1 Answer 1

6 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

6 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related