Leaderboard

Zipf’s law as proposed by American linguist George Kingsley Zipf, states that frequency of any word in a language relates inversely with the rank of that word in frequency table. Frequency is the number of time, the word is appearing in the sample or a text . -The most commonly evident word in a corpus suppose has a frequency f -The second most appearing word , will have a frequency around f/2 -The third most appearing word would have frequency nearly f/3 -The fourth most appearing word would have frequency around f/4 Zipf’s law can be applied in many other data and rankings such as: - Poulation ranks of cities in various countries -Corporation size -Income rankings of richest persons -Ranks or number of people watching the same TV channel -Temprature trends over recent years -Facebook likes of favourite teams - Number of citations to papers - Number of hits on web sites - Copies of books sold in the US - Telephone calls received - Magnitude of earthquakes - Diameter of moon craters - Intensity of solar flares - Intensity of wars - Net worth of Americans -Frequency of family names The level of fit between the data and Zipf’s distribution, can be tested by Kolmogorov-Smirnov test and then it be compared with the fits to alternative distributions, like lognormal, exponential distribution. Zipf’s Law is more or less in compliance with one of the most widely acclaimed economical and statistical principles ‘Pareto Principal’. The Zipf distribution is also labelled as the discrete Pareto distribution, as it includes primarily the discrete data and deals with frequency and rankings. The Pareto principle states that 20% of the invested input accounts for 80% of output. 20 % of work-related input yields 80% of the results. Similarly Zipf’s law accounts for the fact that few of the words, only first 20% of words, accounts for 80% frequency of entire corpus. The probability mass function(pmf) of the Zipf distribution is F(x)= C/xs , C- Calculated Constant, x=1, 2, 3-------------------n S- Value of exponent, characterizing the distribution