Metrics for mass-count disparity

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

20 Scopus citations

Abstract

Mass-count disparity is the technical underpinning of the "mice and elephants" phenomenon - That most samples are small, but a few are huge - which may be the most important attribute of heavy-tailed distributions. We propose to visualize this phenomenon by plotting the conventional distribution and the mass distribution together in the same plot. This then leads to a natural quantification of the effect based on the distance between the two distributions. Such a quantification addresses this important phenomenon directly, taking the full distribution into account, rather than focusing on the mathematical properties of the tail of the distribution. In particular, it shows that the Pareto distribution with tail index 1 < a < 2 actually has a relatively low mass-count disparity; the effects often observed are the result of combining some other distribution with a Pareto tail.

Original languageAmerican English
Title of host publicationProceedings - 14th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2006
Pages61-68
Number of pages8
StatePublished - 2006
Event14th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2006 - Monterey, CA, United States
Duration: 11 Sep 200614 Sep 2006

Publication series

NameProceedings - IEEE Computer Society's Annual International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems, MASCOTS
ISSN (Print)1526-7539

Conference

Conference14th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, MASCOTS 2006
Country/TerritoryUnited States
CityMonterey, CA
Period11/09/0614/09/06

Fingerprint

Dive into the research topics of 'Metrics for mass-count disparity'. Together they form a unique fingerprint.

Cite this