Statalist


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

st: RE: Efficiently run three loops (or do without them)


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Efficiently run three loops (or do without them)
Date   Mon, 13 Apr 2009 13:48:21 +0100

It is difficult to advise how to do something unspecified in an
efficient way. 

Nor do I see why you think you need to throw out most of the data at
each step. 

But if the idea is to check each cross-combination of variables, then
-egen group = group(<varlist>)- creates a single grouping variable. 

Nick 
n.j.cox@durham.ac.uk 

Ronnie Babigumira

I have quarterly data that looks like this

       qtr   houscode   fup_pdt   fup_unit   fup_qtycoll
         1        562        23          2            50
         1        570       628          2             2
         1        573       628        201            10
         1        573       628          2             2
         1        576       628        201             5
         1        576       628        201            20
         1        577       628          2             1
         1        578       628          2             1
         1        590        34         26            60
         1        595        34         26           200

For each quarter, I would like to identify "strange" values (outliers)
in the variable fup_qtycoll
(simply to rule out data entry error).

This would be done for the different fup_pdt and fup_unit combinations

My initial idea is that I would have to do it in three -foreach- loops,
be something along the lines
(this does not work since I need to -preserve- before -keep-ing, which I
would like to avoid, and it
is not by quarter yet)

levelsof fup_pdt, local(fupdts)
foreach i of local fupdts {
		keep if fup_pdt == `i'
		levelsof fup_unit, local(fupunits)
		foreach j of local j {
			keep if fup_unit == `j'
			*** run my checks and other stuff
	    }	
}

I would appreciate some help on how I can do this efficiently

*
*   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/



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