1
$\begingroup$

I have a data set with a range of 0 to 65,000. The vast majority of data points (it is a huge sample) are concentrated between 0 and 1000. There is only one point that has 65,000. I want to plot this using a semi-logarithmic plot. However, I would like the graph to have around 50 points. If I use scales like 2,4,6,8,16,32,64,128,256 etc I have only a couple of points. But If I use a multiple of 1.1 or 1.5 etc, the scales are not integers. Is there a standard way to ensure you more concentrated intervals at a lower level on the scale?

Cheers,

s

$\endgroup$

2 Answers 2

3
$\begingroup$

If you really only have one data point greater than 1000, it would be easiest just to delete that point from the graph. You can make a note in the caption or as a text box that there is an outlier.

$\endgroup$
0
2
$\begingroup$

Can you just use a scale comprised of powers of (1+r) for some small r, and round to the nearest integer? For example, in R, with r = 0.25:

> x <- unique(round(1.25^(0:50)))
> x
 [1]     1     2     3     4     5     6     7     9    12    15    18    23    28    36    44    56    69    87   108   136   169   212   265   331   414
[26]   517   646   808  1010  1262  1578  1972  2465  3081  3852  4815  6019  7523  9404 11755 14694 18367 22959 28699 35873 44842 56052 70065
$\endgroup$
1
  • $\begingroup$ This is a resourceful approach and yet I wonder whether its benefits are outweighed by the problems it might pose for interpretability for some audiences. $\endgroup$
    – rolando2
    Apr 11, 2011 at 16:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.