|
In his paper Kronecker Graphs: An approach to modeling Networks Jure et Al, mention that an important property of networks are that they are heavy tailed. I'm trying to get an insight on what this really means. Do you have good examples of real heavy-tailed distributions, or what does it really means. Thanks |
|
A heavy tailed distribution has thicker tails than an exponential distribution. In practice this often means that it is more difficult to estimate due to a high number of outliers. Good examples of this are distributions that display power law behaviour or the Zipf distribution that describes the frequencies of word frequencies. So there are just a few words that occur very often and many of them that occur just once in a text. Another example is the distribution of city sizes where most cities are tiny but some are so big that they still contain a significant proportion of the total population. This does not apply to all heavy tailed distributions and is kind of the extreme but illustrates how some heavy tail distributions differ from, for example, the Gaussian. |
