[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
Dari Sylvester <darisylvester@yahoo.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
Re: st: Run time for GLLAMM multilevel modeling |

Date |
Mon, 22 Nov 2004 07:10:41 -0800 (PST) |

Michael, Thank you kindly for your suggestions. Best, Dari --- Michael Ingre <Michael.Ingre@ipm.ki.se> wrote: > Dear Dari, > > Unfortunately, -gllamm- relies on numerical > integration and is not very > fast in higher dimensional models, especially not > with continuos > outcomes. Also, it is recommended to use the -adapt- > option with > continuos outcomes. There are two techniques you > could use to speed up > -gllamm-. 1) Use -contract- to reduce the number of > observations and > calculate frequency weights for use with the -wt()- > option of gllamm. > 2) Start with fewer integration points to get > approximate starting > values for use in your final model. Computing time > increases approx > exponentially with number of integration points and > the number of > random effects. The first option might not be very > efficient for you > because you have (it seems) continuos x-variables in > your model. The > second approach looks like this: > > . gllamm volun schmooz soctrust macher happy educ > income marital /// > , i(respno friend zip) nip(4) adapt // starting > values with 4 > quadrature points > . matrix b = e(b) > . gllamm volun schmooz soctrust macher happy educ > income marital /// > , i(respno friend zip) from(b) nip(8) adapt // > increase to 8 > . matrix b = e(b) > . gllamm volun schmooz soctrust macher happy educ > income marital /// > , i(respno friend zip) from(b) nip(16) adapt // > check the stability > of the results see -whelp quadchk- for info. > > Michael > > > > On 2004-11-20, at 13.33, Dari Sylvester wrote: > > > Greetings, Statlisters: > > > > I am currently running a 3-level nested model in > > gllamm with an n=30,000. It has been running for > 39 > > hours and it on iteration 7. > > > > I have been told these programs may run overnight, > but > > never heard of something this long. Since I'm new > to > > gllamm, I'm afraid I've done something wrong. > I've > > included the commands below: > > > > gllamm volun schmooz soctrust macher happy educ > income > > marital, i(respno friend zip) > > > > ---------------------------------------- > > Volun = DV; schmooz - marital = explanatory vars; > > respno, friend, zip are my 3 levels > > > > Does anyone have any insight on this?? > > > > Thanks much! > > Dari > > > > > > > > ===== > > > > <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< > > > > Dari Elise Sylvester > > > > Doctoral Candidate > > > > Department of Political Science, Stony Brook > University > > > >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> > > > > > > > > > > > > > > > > > > __________________________________ > > Do you Yahoo!? > > The all-new My Yahoo! - Get yours free! > > http://my.yahoo.com > > > > > > * > > * For searches and help try: > > * > http://www.stata.com/support/faqs/res/findit.html > > * http://www.stata.com/support/statalist/faq > > * http://www.ats.ucla.edu/stat/stata/ > > > > > ------------------------------------------------ > Michael Ingre , PhD student & Research Associate > Department of Psychology, Stockholm University & > National Institute for Psychosocial Medicine IPM > Box 230, 171 77 Stockholm, Sweden > > * > * For searches and help try: > * > http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > ===== <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< Dari Elise Sylvester Doctoral Candidate Department of Political Science, Stony Brook University >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

**Follow-Ups**:**Re: st: Run time for GLLAMM multilevel modeling***From:*Stas Kolenikov <skolenik@gmail.com>

**References**:**Re: st: Run time for GLLAMM multilevel modeling***From:*Michael Ingre <Michael.Ingre@ipm.ki.se>

- Prev by Date:
**st: testing** - Next by Date:
**st: reproduce bootstrap result** - Previous by thread:
**Re: st: Run time for GLLAMM multilevel modeling** - Next by thread:
**Re: st: Run time for GLLAMM multilevel modeling** - Index(es):

© Copyright 1996–2016 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |