Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm looking for a way to truncate outliers in a time-series graph.

Some context: I'm plotting two different metrics on a single graph. It's important to understand what portion of "Metric A" takes up respective to "Metric B". "Metric B" can at times spike to a high value thus increasing the Y-Axis scale which then minimizes the visibility of Metric "A".

Are there any good formulas that help normalize this data or visual treatments that can make the graphic honest as well as accessible?

See screenshot for more context. The blue area is "Metric B". enter image description here


share|cite|improve this question
up vote 1 down vote accepted

If Metric B spikes up too much (say to more than twice the maximum of Metric A), you can display it as a vertical blue column with a jagged break in the middle to indicated it goes off the top of the graph. I have also seen it with a jagged break and an arrow on the top. You can also list the specific data point to show how tall the spike is.

share|cite|improve this answer
Nice. Makes sense. Is there a rule of thumb on how to determine the threshold? I want to make sure I'm not being dishonest with the data, but still be able to tell the story. – Tom R May 28 '13 at 15:22
@TomR: I don't think there is an agreed threshold. The higher you get, the more Metric A gets compressed. The lower you go, the more common it is to happen. The point is to show that it has run off the paper. I haven't found a nice image to show what I am thinking, but you have probably seen them. – Ross Millikan May 28 '13 at 15:46

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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