Notice: On March 31, it was **announced** that Statalist is moving from an email list to a **forum**. The old list will shut down at the end of May, and its replacement, **statalist.org** is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

From |
Stas Kolenikov <skolenik@gmail.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
Re: st: GAM in survey commands |

Date |
Wed, 24 Feb 2010 15:00:48 -0600 |

My thinking about GAM is that (in the linear regression case) you represent the relation between y and x as y = sum over k of beta_k f_k(x) + error where f_k are the basis functions. In the early days of regression, powers of x were popular, but they have obvious problems at the edges. More typical modern examples are splines. My favorite class of the basis functions are B-splines which are smooth piecewise cubic functions that looks like bumps centered at some points of interest (typically on a grid over the x axis). Type -findit spline- to get a few ideas; I had my own toy routine to create B-splines, a more official-looking one was written by Roger Newson, and I think Stata 11 has something about splines although I have not looked into that. On Wed, Feb 24, 2010 at 2:10 PM, Priscilla Martinez < priscilla.martinez@medisin.uio.no> wrote: > Thanks Stas. I'm fairly new to stata, so I'm not entirely sure what you > mean by "basis functions", but gather that they are functions that modify > the relationship between y and x. Do you know where I could find any > examples of such? > > thx, > P > > > On Wed, Feb 24, 2010 at 6:56 AM, Priscilla Martinez < > > priscilla.martinez@medisin.uio.no> wrote: > > > >> I'm working on a complex survey sample and want to confirm that running > >> a > >> generalized additive model is not feasible with survey commands? I've > >> done > >> numerous searches and it seems so, but I want to make sure I'm not > >> missing > >> something. > >> > > > > Depends on how you approach it. If you are using a 10-year old -gam- > > command > > by Patrick Royston, then the answer is no. If you produce your own basis > > functions and plug them into -regress- or -glm-, then of course that > works > > fine with -svy- prefix. > > > > -- > > Stas Kolenikov, also found at http://stas.kolenikov.name > > Small print: I use this email account for mailing lists only. > > > > * > > * For searches and help try: > > * http://www.stata.com/help.cgi?search > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > * > * For searches and help try: > * http://www.stata.com/help.cgi?search > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > -- Stas Kolenikov, also found at http://stas.kolenikov.name Small print: I use this email account for mailing lists only. * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

**Follow-Ups**:**Re: st: GAM in survey commands***From:*"Priscilla Martinez" <priscilla.martinez@medisin.uio.no>

**References**:**st: GAM in survey commands***From:*"Priscilla Martinez" <priscilla.martinez@medisin.uio.no>

**Re: st: GAM in survey commands***From:*Stas Kolenikov <skolenik@gmail.com>

**Re: st: GAM in survey commands***From:*"Priscilla Martinez" <priscilla.martinez@medisin.uio.no>

- Prev by Date:
**Re: st: RE: Descriptives using multiple-imputed complex survey design data** - Next by Date:
**Re: st: RE: regression r(103): too many variables** - Previous by thread:
**Re: st: GAM in survey commands** - Next by thread:
**Re: st: GAM in survey commands** - Index(es):