**Left Censored » R**, and kindly contributed to R-bloggers)

I’m currently working on a paper that uses Polish survey data (EVS 2008). I am specifically looking at regional variation in particular responses. Because there are only around 1800 observations in the survey, which are split across 66 subregions of Poland (NUTS-3, specifically), I suspected there would be a large degree of variation in how these interviews were distributed across regions. Typically, I would just use `densityplot`

from the `lattice`

package to get some idea of how a continuous variable is distributed. Of course, with discrete data, `table`

would work well when the variable only takes on few possible values. When the variable can take on a larger number of values, `barchart`

(also from `lattice`

) may also work. However, none of these seemed to provide the type of information I wanted. `densityplot`

obscured the distribution of the data, there were too many categories for `table`

to be all that useful, and I found the `barchart`

to be ugly and not that informative. What I came up with was the following (click the image to get the PDF version):

This is a simple variation of a frequency plot (I like simple plots), but I found it to be much more informative than the alternatives. I hope it’s obvious that each dot represents a NUTS-3 region in Poland, the *x*-axis, as the label states, is the number of interviews conducted in each region. The function I used to create this plot is as follows:

distplot <- function(x, ...) { d <- table(x) d <- do.call(rbind, tapply(d, d, function(x) cbind(x, 1:length(x)))) xyplot(d[,2] ~ d[,1], ...) }

And this is how it was called:

distplot(Data$subreg.id, ylab = NULL, xlab = "Number of interviews conducted", scales = list(x = list(at = seq(0, 80, 5)), y = list(at = seq(0, 20, 5))), col = red[7], pch = 16, ylim = c(0,15))

It is quite possible (even likely) that there is a more elegant way to produce a similar plot—there may even be a built-in function somewhere. But sometimes it’s just quicker to code something yourself than spending a bunch of time looking for “a better way”.

**leave a comment**for the author, please follow the link and comment on their blog:

**Left Censored » R**.

R-bloggers.com offers

**daily e-mail updates**about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...