Talk:Normal-inverse Gaussian distribution
|WikiProject Statistics||(Rated Start-class, Low-importance)|
Seriously, you can't understand this article as someone without having knowledge about certain aspects of the topic, this is way too complicated. You should write it so that anyone can understand it. --220.127.116.11 (talk) 21:56, 1 June 2011 (UTC)
Moments of NIG distributions are discussed in, e.g., Hanssen, A. and Oigard, T. A. (2001). The normal inverse Gaussian distribution: a versatile model for heavy-tailed stochastic processes. Acoustics, Speech, and Signal Processing, 2001. Proceedings. 2001 IEEE International Conference on. Volume 6, pp. 3985 - 3988; http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel5/7486/20359/00940717.pdf Excess kurtosis equals to the corrected expression, with 3 subtracted.
[Another user] I don't have access to the mentioned article, but I implemented the computation of the moments and checked with a numerical integration. I disagree with the value of the excess kurtosis: you should not subtract 3. Moreover, this conforms with the implementation I found for Octave: http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=6050&objectType=File despite the fact that this article agrees with you: http://depts.washington.edu/sce2003/Papers/05.pdf —Preceding unsigned comment added by 18.104.22.168 (talk) 16:09, 7 July 2008 (UTC)
[Yet another user] The location of the kurtosis is often shifted by 3 in order to obtain zero kurtosis for the gaussian distribution. Sometimes this adjusted kurtosis is referred to as "excess kurtosis", i.e. gaussian distribution has zero excess kurtosis (mesokurtic). Platykurtic and leptokurtic distributions have negative/positive excess kurtosis. As for sources, you could use "Processes of normal inverse Gaussian type", Finance and stochastics [0949-2984] Barndorff Nielsen yr:1997 vol:2 iss:1 pg:41. —Preceding unsigned comment added by 22.214.171.124 (talk) 10:03, 12 April 2011 (UTC)