0

In mathematics, skew refers to the difference between the mean and the median of the data. This means that A = [1000, 1000, 1000, 1000] and B = [10, 1000, 1000, 10] are both NOT skewed.

In the field of data streaming skew ofter refers to the data not being uniform, so there B, normal distribution, exponential distribution, Zipfian distribution would all be considered a skewed distribution (they have heavy elements).

What is the correct mathematical term to describe that kind of "skew"?

MrAkroMenToS
  • 101
  • In depends in which context. I'm guessing you want in data distributions, which has a very specific definition - https://en.wikipedia.org/wiki/Skewness. Note that the normal distribution has no skew (contrary to what you stated) – Calvin Lin Apr 26 '20 at 15:32
  • 1
    There are other definitions of skewness, but most are related to lack of some sort of symmetry in the distribution. Your later examples are simply non-uniform distributions and you can use that to describe them – Henry Apr 26 '20 at 15:35
  • For me the only thing (researching data streaming) that matters is how far is a distribution from the uniform. Papers I read often use the data's skewness as an indication of how "non-uniform" they are. If I understand I CANNOT use the term "skewed" for this and can only say "distribution is non-uniform"? (Are they also misusing it?) – MrAkroMenToS Apr 26 '20 at 15:54

0 Answers0