This is a difficult task. If Xand Yare bivariate normal then the regression E[Y|X] is linear. If they are bivariate alpha stable the regression is non linear. Have a look at the material on multivariate alph stable distributions in Uchaikin and Zolotarev (1999), Chance and Stability, VSP. There are details of the estimation in Canbanis, Samorodnitsky and Taqqu (1991), Stable Processes and Related Processes, Birkhauser.
In http://www.tcd.ie/Economics/staff/frainj/Stable_Distribution/thesis_main_5.pdf. There is a chapter dealing with the case where the regressors are fixed and the residual is alpha stable. In that case I did not use R as I had considerable problems with fitting the required alpha stable functions in R. You might also look at John P Nolan's alpha stable libraries. There is a version available for R which extends the alpha stable facilities available in R. Best Regards John On Sunday, 14 August 2011, Pascal Grosbuis <pascal.grosb...@orange.fr> wrote: > Hello ! > > I'm already using "fBasics" to generate alpha-stable variables or compute their density or distribution function but > do you know where I could find .R tools for computing the correlation and fit a regression between two > alpha-stable variables ? > > Thanks in advance ! > > Kind regards, > > Pascal Grosbuis > (France) > [[alternative HTML version deleted]] > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > -- John C Frain Economics Department Trinity College Dublin Dublin 2 Ireland www.tcd.ie/Economics/staff/frainj/home.html mailto:fra...@tcd.ie mailto:fra...@gmail.com [[alternative HTML version deleted]] ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
