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

From |
siyama@who.int |

To |
statalist@hsphsun2.harvard.edu |

Subject |
st: RE: RE: Using egen or similar |

Date |
Sat, 23 Aug 2003 16:37:30 +0200 |

Dear Lee, Many thanks for your help, I think I solved the problem with your suggestion to use "egen, concat". Vars m1-m6 are coded 0/1 so I started with the following: for X in num 2/6: recode mX 1=X egen new=concat(m1 m2 m3 m4 m5 m6) and I did get the desired output as follows: new | Freq. Percent Cum. ----------------------------------------------- 000000 | 21 7.07 7.07 100000 | 2 0.67 7.74 103000 | 1 0.34 8.08 103456 | 3 1.01 9.09 123000 | 3 1.01 10.10 123400 | 2 0.67 10.77 123450 | 2 0.67 11.45 123456 | 263 88.55 100.00 ------------+----------------------------------- Total | 297 100.00 Once again, my thanks for your good suggestion and time. Amani -----Original Message----- From: Lee Sieswerda [mailto:Lee.Sieswerda@tbdhu.com] Sent: Saturday, 23 August 2003 00:00 To: 'statalist@hsphsun2.harvard.edu' Subject: st: RE: Using egen or similar I think this will give you pretty close to what you want. * Reduce your master data to only the six * variables of interest, and concatenate * into a new variable use your_master_data.dta keep m1-m6 egen bincats = concat(m1-m6) keep bincats sort bincats save temp1.dta, replace * Generate a new dataset with all * possible categories, in this case * you've got 2^6 possible categories clear set obs 64 gen cat1 = 0 gen cat2 = 0 gen cat3 = 0 * Okay, this block is a hack job. I know * there is a nice algorithm for * generating all of the permutations * of a binary outcome over n events, but * I can't put my hands on it right now replace cat1 = 1 in 33/64 replace cat2 = 1 in 17/32 replace cat2 = 1 in 49/64 replace cat3 = 1 in 9/16 replace cat3 = 1 in 25/32 replace cat3 = 1 in 41/48 replace cat3 = 1 in 57/64 egen cat4 = fill(0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1) egen cat5 = fill(0 0 1 1 0 0 1 1) egen cat6 = fill(0 1 0 1) egen bincats = concat(cat1-cat6) * Now generate the kind of * labels you want, more or less forvalues n = 1/6 { replace cat`n' = cat`n'*`n' } egen numcats = concat(cat1-cat6) keep numcats bincats * Now merge your original data back into these * categories sort bincats merge bincats using temp1.dta * Drop the categories not needed drop if _merge==1 * And tabulate tab numcats There may also be a clever way to do this with string functions rather than with -merge-. Regards, Lee Lee Sieswerda, Epidemiologist Thunder Bay District Health Unit Lee.Sieswerda@tbdhu.com -----Original Message----- From: siyama@who.int [mailto:siyama@who.int] Sent: Friday, August 22, 2003 2:43 PM To: statalist@hsphsun2.harvard.edu Subject: st: Using egen or similar Dear Stata-listers, I have six binary variables (m1,..,m6) coded 0/1 I wish to see the frequency of the pattern of occurring 1s, but preserving the variable sequence. I am not sure whether any of "egen" functions can perform the task (and I prefer not to use egen..=group(..)). I started off with the following: for X in num 2/6: recode mX 1=X then if there is one of egen functions it would be egen new=fun(m1 m2 m3 m4 m5 m6) then tabulate new would yield new freq ------ ------ 1,2,3 63 1,3,4,5 23 1,4,5 30 2,5,6 20 .....etc It is if I am "xtdes" but with variables rather than panel time-entries. Many thanks for your help in advance. Amani * * 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/ * * 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/

- Prev by Date:
**Re: st: Trouble with Win XP/Office XP** - Next by Date:
**Re: st: SP/RM[1,2] EMS** - Previous by thread:
**st: Confidence Interval for Predicted Probability using end points** - Next by thread:
**st: prtab** - Index(es):

© Copyright 1996–2015 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |