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

From |
"Dupont, William" <william.dupont@vanderbilt.edu> |

To |
<statalist@hsphsun2.harvard.edu> |

Subject |
st: RE: RE: Graph matrix with non-linear regression lines |

Date |
Thu, 12 Jun 2003 14:25:11 -0500 |

Nick I like your program. Was it your idea to use the diagonal for percentile plots? If so it is a clever use of this space. Bill -----Original Message----- From: Nick Cox [mailto:n.j.cox@durham.ac.uk] Sent: Thursday, June 12, 2003 1:21 PM To: statalist@hsphsun2.harvard.edu Subject: st: RE: Graph matrix with non-linear regression lines Bill Dupont > In version 7, the -matrix- and -band- options of the graph > command gave > matrix graphs with a non-linear regression line added to > each scatter > plot. These regression lines appear to be unavailable in > the version 8 > graph matrix command. Assuming I have not misread the > manual, what are > listers thoughts on the utility of these regression lines on matrix > plots? What were Stata's reasons for dropping this option? > I thought it was useful. -band()- showed conditional group medians. That's, I guess, a kind of L_1 regression. As you're polling, I hadn't noticed it had gone, despite using -graph matrix- quite a lot, so that's lukewarm at best. Why has it gone is a question for Stata Corp, but I note that -graph matrix- is in essence a matrix of -scatter-. In principle, there are all sorts of graphs which can be put into a matrix, and while we're waiting for Stata Corp to give us some easy handles for that, a challenge is to see how far one can get. Below my signature is an example. This is just -my- -graph- -matrix- on 12 June. I decided to put quantile plots on the diagonal, and I prefer -lowess-, but the idea can be varied fairly easily. Nick n.j.cox@durham.ac.uk P.S. it's slow. (OK, now don't start that one again.) program mygraphmatrix *! NJC 1.0.0 12 June 2003 version 8 syntax varlist(numeric) [if] [in] [, * ] marksample touse qui count if `touse' if r(N) == 0 error 2000 set graphics off qui foreach y of local varlist { local ny = 1 foreach x of local varlist { tempname f if `ny' > 1 { local yt } else { local yt : variable label `y' if `"`yt'"' == "" local yt "`y'" } if "`x'" == "`y'" { quantile `x', yti("`yt'") saving(`f'.gph) } else { lowess `y' `x', title("") note("") /// legend(off) /// yti(`"`yt'"') saving(`f'.gph) `options' } local files `"`files' "`f'""' local FILES "`FILES' `f'.gph" local ny = `ny' + 1 } } set graphics on graph combine `files', imargin(zero) foreach F in `FILES' { erase `F' } end * * 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/ * * 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**:**st: RE: RE: RE: Graph matrix with non-linear regression lines***From:*"Nick Cox" <n.j.cox@durham.ac.uk>

- Prev by Date:
**Re: st: meta-analysis for diagnostic test** - Next by Date:
**st: RE: RE: RE: Graph matrix with non-linear regression lines** - Previous by thread:
**st: meta-analysis for diagnostic test** - Next by thread:
**st: RE: RE: RE: Graph matrix with non-linear regression lines** - Index(es):

© Copyright 1996–2014 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |