CODEWITHSUNDEEP
rgraphicsdata-visualization
PoisonAlien
PoisonAlien

Reputation: 431

best visualization method

I have data which looks like this: 

     chr01 chr02 chr03 chr04 chr05 chr06 chr07 chr08 chr09
T10     2     5     3     5     4     1     9     2     3
T11     0     2     1     5     2     1     3     5     4
T65     0     5     1     3     4     1     5     3     1

Some of the columns columns have 0's. I want to visualize the number of zeros in each column (may be as a percentage content of 0 for each column). I am an R user, first I thought of using pie chart but I was wondering is there any sophisticated way of representing it !? Even I tried heatmap. Any other way to represent this?? (bottom line is I want to represent percentage of 0's per column wise)

Upvotes: 0

Views: 256

Answers (3)

Geek On Acid
Geek On Acid

Reputation: 6410

Another way is to use dotplot - where you only represent your values by a single dot. I would use package lattice to do this instead of ggplot2, but I added both solutions below just in case:

#load df data from @Arun answer, and then...
library(reshape2)#for melt function
dd <- apply(df,2,function(x) mean(x==0)*100)
d1 <- melt(dd)#gets data to long format
d <- data.frame(variable=rownames(d1), d1) 
#lattice dotplot
library(lattice)
dotplot(reorder(variable, value) ~ value, d, col=1, aspect=1, 
        xlab="percentage", ylab="")

enter image description here

#ggplot2 dotplot 
library(ggplot2)
ggplot(d, aes(x = value, y = reorder(variable, value))) + 
  labs(x = "percentage", y = "") +
  geom_point() + 
  theme_bw()

enter image description here

Upvotes: 2

Arun
Arun

Reputation: 118789

You could also use ggplot2. It gives you more control, although I am not sure if that's the eye-candy you're looking for. I am not sure if you are asking for a completely different type of visualisation or if you're looking for plotting a bar-plot (which is appropriate here as @Didzis showed) with more control. For the 2nd case, ggplot2 might be useful:

require(ggplot2)
df <- structure(list(chr01 = c(2L, 0L, 0L), chr02 = c(5L, 0L, 5L), 
         chr03 = c(3L, 1L, 0L), chr04 = c(0L, 5L, 0L), chr05 = c(0L, 
         2L, 4L), chr06 = c(0L, 0L, 0L), chr07 = c(9L, 3L, 0L), chr08 = c(2L, 
         0L, 3L), chr09 = c(3L, 4L, 1L)), .Names = c("chr01", "chr02", 
         "chr03", "chr04", "chr05", "chr06", "chr07", "chr08", "chr09"
         ), class = "data.frame", row.names = c("T10", "T11", "T65"))
gg.df <- data.frame(chr.id = names(df))
gg.df$cnt <- sapply(df, function(x) sum(x==0)/length(x) * 100)

qplot(factor(chr.id), weight=cnt, data=gg.df, geom="bar", fill=factor(chr.id))

This gives you: ggplot2 bar plot example.
Of course you can change every element of this plot (check out the link at the beginning of this post).

Upvotes: 1

Didzis Elferts
Didzis Elferts

Reputation: 98429

Simple way to represent results is to make bar plot. Assuming that your data frame is named df:

#Calculate percentage of 0 for each column
pr.0<-apply(df,2,function(x) mean(x==0)*100)
#Plot results
barplot(pr.0,ylab="Percentage")

enter image description here

Upvotes: 1

Related Questions