Skip to content

barplot that displays sums of values of 2 y colums grouped by different variables

5 messages · Eric Berger, Jim Lemon, Jeff Newmiller +1 more

Original
kenneth dyson
# 
I am trying to create a barplot displaying the sums of 2 columns of data 
grouped by a variable. the data is set up like this:

"city" "n" "y" <br>
mon 100 200 <br>
tor 209 300 <br>
edm 98 87 <br>
mon 20 76 <br>
tor 50 96 <br>
edm 62 27 <br>

the resulting plot should have city as the x-axis, 2 bars per city, 1 
representing the sum of "n" in that city, the other the sum of "y" in 
that city.

If possible also show the sum in each bar as a label?

I aggregated the data into sums like this:

sum_data <- aggregate(. ~ City,data=raw_data,sum)

this gave me the sums per city as I wanted but for some reason 1 of the 
cities is missing in the output.

Using this code for the plot:

ggplot(sum_data,aes(x = City,y = n)) + geom_bar(aes(fill = y),stat = 
"identity",position = "dodge")

gave be a bar plot with one bar per city showing the sum of y as a color 
gradient. not what I expected given the "dodge" command in geom_bar.

Thanks.
Eric Berger
# 
'position="dodge"' has no effect in the plot because the x-axis is a factor
variable. The bars do not need to be moved to avoid each other. The
'aes(fill=y)' is specifying that you want the color gradient to capture the
sums in the 'y' variable. You might be better off to use 'no' and 'yes'
rather than 'n' and 'y' to avoid confusion. Then you would see that the
statement would be 'aes(fill=yes)'. Summary: the height of each bar
represents the sum of the 'no' for that city, and the color of each bar
represents the sum of the 'yes' for that city. Your code is fine, unless
that is not what you were trying to do.

HTH,
Eric


On Mon, Jan 15, 2018 at 6:59 PM, kenneth dyson <kenneth at kidscodejeunesse.org

  
  


  


      

    

    

      
      

        


  

        

      Jim Lemon
    

    

      
      #
    

  


  

      
Hi Kenneth,
I don't know about ggplot, but perhaps this will help:

kddf<-read.table(text="city n y
mon 100 200
tor 209 300
edm 98 87
mon 20 76
tor 50 96
edm 62 27",
header=TRUE,stringsAsFactors=FALSE)
library(plotrix)
barpos<-barp(t(kddf[,2:3]),names.arg=kddf[,1],xlab="City",ylab="Sum",
 main="Sums of values of 2 y columns",col=2:3)
legend(5,300,c("Sum of n","Sum, of y"),fill=2:3)
barlabels(barpos$x,barpos$y)

Jim


On Tue, Jan 16, 2018 at 3:59 AM, kenneth dyson

        
<kenneth at kidscodejeunesse.org> wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Jeff Newmiller
    

    

      
      #
    

  


  

      
It is not generally advisable to get too fancy with stat functions in 
ggplot... things can easily get more complicated than ggplot is ready to 
handle when it comes to calculations. It is better to create data that 
corresponds directly to the graphical representations you are mapping 
them to.

Read [1] for more on this philosophy.

[1] H. Wickham, Tidy Data, Journal of Statistical Software, vol. 59, no. 
10, pp. 123, Sep. 2014. http://www.jstatsoft.org/v59/i10/

#---
library(ggplot2) # ggplot
library(dplyr)   # `%>%`, group_by, summarise
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#>
#>     filter, lag
#> The following objects are masked from 'package:base':
#>
#>     intersect, setdiff, setequal, union
library(tidyr)   # gather

dta <- read.table( text =
"city n y
mon 100 200
tor 209 300
edm 98 87
mon 20 76
tor 50 96
edm 62 27
", header = TRUE )

dta2 <- (   dta
         %>% group_by( city )
         %>% summarise( n = sum( n )
                      , y = sum( y )
                      )
         %>% gather( Response, value, -city )
         )

ggplot( dta2, aes( x=city, y=value, fill = Response ) ) +
     geom_bar( stat="identity", position="dodge" )

#' ![](https://i.imgur.com/cosFf3B.png)
#---

        
On Mon, 15 Jan 2018, kenneth dyson wrote:

        

            

      
      
      
    

        

      
      
    

        
---------------------------------------------------------------------------
Jeff Newmiller                        The     .....       .....  Go Live...
DCN:<jdnewmil at dcn.davis.ca.us>        Basics: ##.#.       ##.#.  Live Go...
                                       Live:   OO#.. Dead: OO#..  Playing
Research Engineer (Solar/Batteries            O.O#.       #.O#.  with
/Software/Embedded Controllers)               .OO#.       .OO#.  rocks...1k

  
  


  


      

    

    

      
      

        


  

        

      kenneth dyson
    

    

      
      #
    

  


  

      
Thanks everyone.

Got it to work like this, if anyone is interested:

import the data with readr, taking in only the columns that have numeric 
values ("n" and "y") and the column with the groups ("city").

aggregate the data by the group ("city") so that each variable has a sum:

|sum_data <-aggregate(.~City__c,data=raw_data,sum)|

reshape the data so that the groups are now variables (there will be 3 
columns: "city", "variable", "value"):

|library(reshape2)library(ggplot2)sums <-melt(sum_data)|

plot using ggplot:

|ggplot(sums,aes(x =city,y=value,fill =variable,ymax 
=1000))+geom_bar(stat="identity",width=.8,position ="dodge")|

        
On 2018-01-15 6:01 PM, Jeff Newmiller wrote: