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

From |
"Nick Cox" <n.j.cox@durham.ac.uk> |

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

Subject |
st: RE: loop |

Date |
Sun, 9 Feb 2003 18:28:19 -0000 |

Jun Xu posted twice, here labelled (1) and (2): (1) > puzzled by a problem when writing an ado file. Suppose I have > var1 var2 var3 var4 var5......vark, and I want to do the > following loop: > ************************************************************ > ********* > var1 > var2 > var3 > ... > ... > ... > vark > var1 var2 > var1 var3 > var1 var4 > ... > ... > var1 vark > var2 var3 > var2 var4 > ... > ... > var(k-1) vark > ... > ... > var1 var2 var3 > var1 var2 var4 > var1 var2 var5 > ... > var1 var2 var3....vark > ****************************************************************** > Basically, what I want to do is like step wise exhausting > all combinations > in a systematic way from univariate, bivariate, trivariate, to > multivariate....Or, I can say for every variable in the > variable list, there > is indicator variable associated with it. I either take > this variable in or > out in each run. And there should be 2^k possibilities. I > have no idea how > to handle that. COuld anyone give me some hint? Many > thanks in advance. (2) I think I might not have explained my problems clearly. I have k indicator variables (coded as 1 or 0) and I would like to know the response patterns (for example for latent class analsis) to these k variables. For example, var1 var2 var3 ....vark 1 0 0 0 0 1 0 0 ... 1 1 0 0 ... ... ... 1 1 1 1 I would like to know for each response pattern, how many cases are there, and programmed into an ado file. My key problem here is how to run through all the combinations (univariate, bivariate, and trivariate) One posibility is that I used the following cods (or reviced version to fit into an ado file) ****************************************************************** clear for num 1/6: set obs 100\ gen xX=invnorm(uniform()) \ gen DxX=xX>0.6 gen pattern=0 local i=1 while `i'<6 { replace pattern=pattern+Dx`i'*10^(6-`i') local i=`i'+1 } aorder list Dx1-Dx6 pattern sort pattern list pattern gen count=1 collapse (sum) count, by(pattern) *********************************************************** The resulting data matrix looks like: ============================ pattern count 0 16 10 10 100 5 110 6 1000 8 1010 7 1100 2 1110 1 10000 11 10010 3 10100 2 10110 2 11000 2 11010 2 11100 1 100000 7 100010 1 100100 2 101000 4 101010 1 110000 1 110010 2 110100 2 111000 1 111010 1 ================================= Here the problem is that it only presents the response pattern that has at least one case and it's hard to handle its order (now is list in numerical order: from small to big) But what if I want to go through "each" combination (2^k possible ways) in a sysmatic way and list all response pattern freqeuncy though some of them have zero cases. What I meant by a systematic way is like: ********************************************************************* var1 var2 var3 ... ... ... vark var1 var2 var1 var3 var1 var4 ... ... var1 vark var2 var3 var2 var4 ... ... var(k-1) vark ... ... var1 var2 var3 var1 var2 var4 var1 var2 var5 ... var1 var2 var3....vark ****************************************************************** or in binary coding **************************************************************** 1 0 0 0 0 .....0 0 1 0 0 0 .....0 0 0 1 0 0 .....0 ... ... ... 0 0 0 0 0 .....1 1 1 0 0 0 .....0 1 0 1 0 0 .....0 1 0 0 1 0 .....0 ... ... 1 1 1 0 0 .....0 1 0 1 1 0 .....0 ... ... ... ... 1 1 1 1 1 .....1 *********************************************** Here I didn't present some summarize command that could grab case number for that response pattern. But basically I will run through each combination and calculate the frequency for that particular combination though there might be zero cases. Thanks a lot >>> To get a tabulation of patterns with some instances, egen all = concat(var1-vark) tab all The following program suggests some possible lines of attack. program permlist, rclass version 8 syntax varlist tokenize `varlist' local nvars : word count `varlist' local imax = 2^`nvars' - 1 forval i = 1 / `imax' { qui inbase 2 `i' local which : di %0`nvars'.0f `r(base)' local vars forval j = 1 / `nvars' { local char = substr("`which'",`j',1) if `char' { local vars "`vars'``j'' " } } local vlist `"`vlist'"`vars'" "' } local varlist forval i = 1 / `nvars' { foreach w of local vlist { local nv : word count `w' if `i' == `nv' { local varlist `"`varlist'"`w'" "' } } } return local varlist `"`varlist'"' end I use the undocumented -inbase- command to get the binary equivalent of 1 ... 2^k - 1 (I omit the null case in which none of the variables are chosen). Nick n.j.cox@durham.ac.uk * * 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: loop (please ignore previous)***From:*"Nick Cox" <n.j.cox@durham.ac.uk>

**References**:**st: loop***From:*"Jun Xu" <mystata@hotmail.com>

- Prev by Date:
**st: infile - multiple lines** - Next by Date:
**st: RE: loop (please ignore previous)** - Previous by thread:
**st: loop** - Next by thread:
**st: RE: loop (please ignore previous)** - Index(es):

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