14

I'm using Orange Canvas and I generated a scatter plot. I have the possibility to jitter continuous variables but I really don't know why would I do that.

What's the idea behind jittering?

amoeba
  • 93,463
  • 28
  • 275
  • 317
Pierre
  • 412
  • 5
  • 16
  • 1
    To deal with overprinting and hence give a less misleading visual impression. If you only have a few points you do not need to do it. – mdewey Dec 23 '16 at 12:36
  • For a concrete example of jittering helping to see the patterns in data, see my answer here: [How to extract information from a scatterplot matrix when you have large N, discrete data, & many variables?](http://stats.stackexchange.com/a/144741/7290) – gung - Reinstate Monica Dec 23 '16 at 17:29

1 Answers1

18

This post explains it well.

Often when you have a lot of overlapping points it can be hard to get a sense for their density. Two thousand points at (1, 1) appear the same as one point at (1, 1). Adding some jitter turns overlapping points into a smear and makes the density of points obvious.

Dex Groves
  • 1,593
  • 8
  • 12
  • 1
    Isn't it about discrete data, whereas the question explicitly asks about continuous data? – amoeba Dec 25 '16 at 12:42
  • 2
    Adding to what @amoeba said this answer is correct when you add that that the continuous data we deal with is rounded due to measurement precision and so problems like described above occur. – Tim Dec 25 '16 at 23:21