On Average, Mean, Median, and Mode

By Daniel Miessler on June 28th, 2010: Tagged as Mathematics
  • CarlM

    Note that distributions may be multi-modal.

  • http://twitter.com/sgharms Steven G. Harms

    I just checked out this book:

    Calculated risks : how to know when numbers deceive you / Gerd Gigerenzer.

    This may be applicable to this topic.

  • http://danielmiessler.com/ Daniel Miessler

    What part do you think is suspect?

  • igf1

    The avg algo that I find most useful in determining mean throughput, is one where I identify abnormalities and remove them. This gives a truer mean. Usually I would have several controls prior to commit, this is important as “abnormality” is very subjective and based on factors outside of the scope of control (which is the only reason to unseat abnormalities in a set). In the pre-samples I drop the highest and lowest and store the output of the remaining mean. Then I average those into a fairly consistent range/median (deviation < .25n). The really interesting part is that the extremes which were cast off, contain important data; through macro analysis.


Top

Popular

Information Security / Technology

Politics

Philosophy & Religion

Technology & Science

Culture & Society

Miscellaneous

Arguments

Projects

Collections

Twitter

What I'm Reading

Favorite Books and Essays

Top Blog Categories

Inputs