Roberto Mura
> I have been working on a do file, trying to re-rank some 60
> variables I have in my
> database. My problem is that i cannot make my loop a
> "smart" loop so to
> speak. My re-ranking do file (is probably very inefficient
> because i am
> relatively new to Stata) is based on pair-wise differences
> among variables.
> So I need Stata to keep on performing the series of
> programs I wrote until
> the condition that none of these 59 differences is
> negative. I have been
> able to create a "dumb loop" like: local i=1 while `i' <=23
> { set more off
> quietly zero quietly primo quietly secondo finale_bis
> display `i' local
> i=`i'+1 }
>
> it takes 23 iterations until all differences are >=0
>
> What I would need is for stata to do something like
>
> while diff1<0 | diff2<0 | diff3<0 | diff4<0 | diff5<0
> |diff6<0 |diff7<0
> |diff8<0 |diff9<0 |diff10<0 | diff11<0 | diff12<0 |
> diff13<0 | diff14<0 |
> diff15<0 |diff16<0 |diff17<0 |diff18<0 |diff19<0 |diff20<0
> | diff21<0 |
> diff22<0 | diff23<0 | diff24<0 | diff25<0 |diff26<0
> |diff27<0 |diff28<0
> |diff29<0 |diff30<0 | diff31<0 | diff32<0 | diff33<0 |
> diff34<0 | diff35<0
> |diff36<0 |diff37<0 |diff38<0 |diff39<0 |diff40<0 |
> diff41<0 | diff42<0 |
> diff43<0 | diff44<0 | diff45<0 |diff46<0 |diff47<0
> |diff48<0 |diff49<0
> |diff50<0 | diff51<0 | diff52<0 | diff53<0 | diff54<0 |
> diff55<0 |diff56<0
> |diff57<0 |diff58<0 |diff59<0 { set more off quietly zero
> quietly primo
> quietly secondo finale_bis }
>
> Now, for some reason I cannot explain myself, it works for only 9
> iterations and then it stops. So it is not "understanding"
> the if condition
> properly.This statement is reinforced by the fact that if I
> run this when
> only, say, 3 more iterations are needed it does nothing as
> after the first
> nine.
>
> The third and last way I could come up with was to
> "encapsulate" that huge
> if condition into a single variable......
>
> by firm: gen elio=1 replace elio=0 if diff1<0 | diff2<0 |
> diff3<0 | diff4<0
> | diff5<0 |diff6<0 |diff7<0 |diff8<0 |diff9<0 |diff10<0 |
> diff11<0 |
> diff12<0 | diff13<0 | diff14<0 | diff15<0 |diff16<0
> |diff17<0 |diff18<0
> |diff19<0 |diff20<0 | diff21<0 | diff22<0 | diff23<0 |
> diff24<0 | diff25<0
> |diff26<0 |diff27<0 |diff28<0 |diff29<0 |diff30<0 |
> diff31<0 | diff32<0 |
> diff33<0 | diff34<0 | diff35<0 |diff36<0 |diff37<0
> |diff38<0 |diff39<0
> |diff40<0 | diff41<0 | diff42<0 | diff43<0 | diff44<0 |
> diff45<0 |diff46<0
> |diff47<0 |diff48<0 |diff49<0 |diff50<0 | diff51<0 |
> diff52<0 | diff53<0 |
> diff54<0 | diff55<0 |diff56<0 |diff57<0 |diff58<0 |diff59<0
>
> while elio ==1 { quietly drop elio set more off quietly
> zero quietly primo
> quietly secondo finale_bis by firm: gen elio=1 replace
> elio=0 if diff1<0 |
> diff2<0 | diff3<0 | diff4<0 | diff5<0 |diff6<0 |diff7<0
> |diff8<0 |diff9<0
> |diff10<0 | diff11<0 | diff12<0 | diff13<0 | diff14<0 |
> diff15<0 |diff16<0
> |diff17<0 |diff18<0 |diff19<0 |diff20<0 | diff21<0 |
> diff22<0 | diff23<0 |
> diff24<0 | diff25<0 |diff26<0 |diff27<0 |diff28<0 |diff29<0
> |diff30<0 |
> diff31<0 | diff32<0 | diff33<0 | diff34<0 | diff35<0
> |diff36<0 |diff37<0
> |diff38<0 |diff39<0 |diff40<0 | diff41<0 | diff42<0 |
> diff43<0 | diff44<0 |
> diff45<0 |diff46<0 |diff47<0 |diff48<0 |diff49<0 |diff50<0
> | diff51<0 |
> diff52<0 | diff53<0 | diff54<0 | diff55<0 |diff56<0
> |diff57<0 |diff58<0
> |diff59<0 }
>
> but that's where I get into a never ending loop! finale_bis
> produces the
> number of differences (count if.....) still negative but,
> after reaching
> the zero, it keeps on performing the operations.
If the diff* are variables then Stata can only
interpret this as the equivalent of
while diff1[1] < 0 ... {
and it may well be that one bug lurks therein.
The issue here is parallel to that for -if-
documented at
http://www.stata.com/support/faqs/lang/ifqualifier.html
Apart from that, it might help you to give us the
whole of your problem, not part of it.
In general terms I would recommend something
more like
local go = 1
while `go' {
do calculations
check if calculations have succeeded
if so {
local go = 0
}
}
Your check should be able to include
a loop over variables. In essence, you should
_never_ need to type something like the compound
condition used here.
Nick
[email protected]
*
* 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/
