I attached an example of what I mean. This plot was generated manually. The algorithm should determine which points can be labelled (where is enough space to add a label; this will be typically for the "outer" points), possibly up to a given maximum number of labels. The labels should then be placed so that neither data points nor other labels will be obscured.
The labelling should be done automatically (i have to generate really many of such plots, with many 100s of points of which the points at the outer range should be labelled)
I know the "work-arounds" to plot the labels without the points and uses a smaller font size, so that overplotting is less likely. But this is not applicable for me. The points must be visible and they are of a particular size. Also, the problem remains how to determine the points that can be labelled (so that the labels remain readable and do not overlap each other).
I hope that someone already solved this problem (what I consider not too exotic, though...). Thanks for hints and helps!
Mohammad, identify() is manual and not automatic. I am looking for an automatic way!
Yes, but with this function you do all you want in R.
For fully automatic procedure, check wordcloud package
In this page, check the first URL, go down until you see the scatter plot, read it, it may help you !
http://cran.r-project.org/web/packages/wordcloud/index.html
If you use lattice or ggplot2 for plot creation, you can use library(directlabels) to label points (or lines or anything really) automatically. directlabels has numerous algorithms to calculate best positioning.
Cheers
Tim
The "thigmophobe.labels" function from package "plotrix" also does the trick.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package