|
Dear all, Right now, I'm doing research about outlier in statistical data (univariate and multivariate) and I want to simulate its behavior. My problem is : How to generate random data from distribution with zero skewness and some kurtosis values in Matlab ? A. Kudus ========================== Dept. of Statistics Bandung Islamic University I n d o n e s i a ========================== |
|
|
Generate rand. data with zero skewness and some kurtosis
Started by ●September 28, 2001
Reply by ●October 8, 20012001-10-08
|
Hi, You can generate the following distribution: P=rand(1,10000); %sample size 10,000 P1=P(P<0.5); P2=P(P>=0.5); X1=(2*P1).^(1/3)-1; X2=1-(2*P2-1).^(1/3); X=[X1 X2]; hist(X,40) % see the distribution Var=sum(X.^2)/1000 % 1000 factor to normalize variance % see properties: sum(X.^3)/1000 ans = 0.0057 % This is zero sum(X.^4)/1000 ans = 0.3016 % This is small Joe is back from vacation -- BSTEX-equation viewer for MATLAB http://www.geocities.com/bstex2001 --- Abdul Kudus <> wrote: > Dear all, > > Right now, I'm doing research about outlier in > statistical data > (univariate and multivariate) and I want to simulate > its behavior. My > problem is : > How to generate random data from distribution with > zero skewness and > some kurtosis values in Matlab ? > > A. Kudus > ========================== > Dept. of Statistics > Bandung Islamic University > I n d o n e s i a > ========================== > > __________________________________________________ |
