boxplot of vectors with different length

问题

I have a matrix of 2 columns. I would like boxplot each of these columns but each has different number of entries.

For example, first column has 10 entries and the second column has 7 entries. The remaining 3 of the second column is given zero.

I would like to plot these side by side for comparison reasons.

Is there a way to tell R to boxplot the whole column 1 and only the first 7 entry for column 2?

回答1:

You could simply index the values you want, for example

## dummy version of your data
mat <- matrix(c(1:17, rep(0, 3)), ncol = 2)

## create object suitable for plotting with boxplot
## I.e. convert to melted or long format
df <- data.frame(values = mat[1:17],
                 vars = rep(c("Col1","Col2"), times = c(10,7)))

## draw the boxplot
boxplot(values ~ vars, data = df)

In the above I'm taking you at your word that you have a matrix. If you actually have a data frame then you would need

df <- data.frame(values = c(mat[,1], mat[1:7, 2]),
                 vars = rep(c("Col1","Col2"), times = c(10,7)))

and I assume that the data in the two columns are comparable in that the fact that the values are in two columns suggests a categorical variable that allows us to split the values (like Height of men and women, with sex as the categorical value).

The resulting boxplot is shown below

回答2:

For any number of columns and any number of empty entries you can do like this.

## Load data from CSV; first row contains column headers
dat <- read.csv( 'your-filename.csv', header = T )

## Set plot region (when set 'ylim' skip first row with headers)
plot(
  1, 1, 
  xlim=c(1,ncol(dat)), ylim=range(dat[-1,], na.rm=T), 
  xaxt='n', xlab='', ylab=''
)
axis(1, labels=colnames(dat), at=1:ncol(dat))

for(i in 1:ncol(dat)) {
  ## Get i-th column
  p <- dat[,i]

  ## Remove 0 values from column
  p <- p[! p %in% 0]
  ## Instead of 0 you can use any values
  ## For example, you can remove 1, 2, 3
  ##   p <- p[! p %in% c(1,2,3)]

  ## Draw boxplot
  boxplot(p, add=T, at=i)
}

This code loads table form CSV files, remove 0 values from the column (or you can remove any other values), and draw all boxplot for every column in one graphic.

Thinks this helps.

来源：https://stackoverflow.com/questions/17749212/boxplot-of-vectors-with-different-length

标签

boxplot