How do I visualize long tails of factor levels?

Question

I have data with some factors, say, a list of music albums by different artists which are owned by me. I counted the unique factors to arrive at a list of artist names with an associated count of albums, like this:

name                             count

Johnny Cash                       10
The Beatles                        7
Blur                               6
Gorillaz                           3
The Good, the Bad and the Queen    1
The Rolling Stones                 1
The Doors                          1
The Jimi Hendrix Experience        1
The Ramones                        1
The Clash                          1
The Who                            1
The White Stripes                  1
The Black Keys                     1
The Specials                       1
The Cure                           1
The Band                           1
The The                            1

What's the best way to visualize that a large portion (one third, to be precise) of my music collection consists of albums of singleton artists?

My first idea was to plot the count as x-value and the cumulative sum of the count as y-value in a way that the y-value increases, similar to a Lorenz plot. I am wondering - are there better ways to visualize this large portion of low-frequencies?

I'd appreciate either a general suggestion to plot what against what or corresponding R-code.

Answer 1

This sounds like a job for Cleveland's dotplot - which is really just a nice looking horizontal barchart.

Assuming your data frame is called df:

R> ## count the number of artist with each number of counts
R> album_counts = table(df$count)
R> ## rev because I find it looks better to have the larger counts at the top
R> dotchart(rev(album_counts))

Another option is to just use a plain old barchart with the barplot() function.

You also allude to the need to compare record collections. If you want to compare several record collections you might just calculate a diversity index for each collection and directly compare those.