I?m trying to find datasets that will give me residuals, after applying the lm function, with no normality, non linearity, and heteroscedacity so I can try to exemplify those cases in the linear regression model. Can you give any advice on what datasets would be appropiate? I can?t use the ones in the alr3 package because those have already been seen in class. Thank you very much :-) natorro -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
Carlos L?pez wrote:> I?m trying to find datasets that will give me residuals, after applying > the lm function, with no normality, non linearity, and heteroscedacity > so I can try to exemplify > those cases in the linear regression model. Can you give any advice on > what datasets would be appropiate? I can?t use the ones in the alr3 > package because those have > already been seen in class. > > Thank you very much :-) > natorro >if you know what you are looking for (or not looking for), wouldn't it be the easiest and fastest thing to do to simulate such a dataset yourself? Best, Roland
Carlos, There are many sources of real datasets (in R itself, on the web), you just need to look a little. For teaching purposes, I think it is always better to use real datasets than to use simulated ones. One thing bothers me, though. You imply that in all the examples you have the data are well fit with linear models, the residuals are normal and there is no sign of heteroscedacity. That sounds a very unusual set of examples! Best Antony> From: Roland Rau <roland.rproject@gmail.com> > Date: 30 May 2008 12:23:17 AM GMT+02:00 > To: Carlos López <natorro@fisica.unam.mx> > Cc: r-help@r-project.org > Subject: Re: [R] Datasets in R > > > Carlos López wrote: >> I´m trying to find datasets that will give me residuals, after >> applying the lm function, with no normality, non linearity, and >> heteroscedacity so I can try to exemplify >> those cases in the linear regression model. Can you give any advice >> on what datasets would be appropiate? I can´t use the ones in the >> alr3 package because those have >> already been seen in class. >> Thank you very much :-) >> natorro > if you know what you are looking for (or not looking for), wouldn't > it be the easiest and fastest thing to do to simulate such a dataset > yourself? > > Best, > Roland[[alternative HTML version deleted]]
Hi Carlos, Carlos L?pez wrote:> I?m trying to find datasets that will give me residuals, after applying > the lm function, with no normality, non linearity, and heteroscedacity > so I can try to exemplify > those cases in the linear regression model. Can you give any advice on > what datasets would be appropiate? I can?t use the ones in the alr3 > package because those have > already been seen in class. > > Thank you very much :-) > natorro >if you don't want to simulate your own data, you might have a look at the NIST Reference Datasets http://www.itl.nist.gov/div898/strd/ I hope this help? Roland