Anywatch

[Ankita Shukla]

Hi, my data is sorted by caseid, year and episodes. I want to find the first non zero value in variable event by caseid, year. If the data was in wide format anymatch command could have solved the purpose but since it is in long form. I am not sure how to solve it. Also I am not very acquainted of looping in stata. If anybody have any ideas about it please share. Thanks!! The format of my data is given below

caseid	year	episode	event
1100122	1	1	0
1100122	1	2	B
1100122	2	1	B
1100122	2	2	0
1100122	2	3	B

[Joseph Coveney]

I'm not quite sure of what it is you're after, but maybe something like the following will give you some ideas to pursue.

.ÿversionÿ15.1

.ÿ
.ÿclearÿ*

.ÿinputÿstr7ÿcaseidÿbyteÿ(yearÿepisode)ÿstr1ÿevent

ÿÿÿÿÿÿÿÿcaseidÿÿÿÿÿÿyearÿÿÿepisodeÿÿÿÿÿÿevent
ÿÿ1.ÿ1100122ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿ0
ÿÿ2.ÿ1100122ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿB
ÿÿ3.ÿ1100122ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿB
ÿÿ4.ÿ1100122ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿ0
ÿÿ5.ÿ1100122ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿ3ÿÿÿÿÿÿÿB
ÿÿ6.ÿend

.ÿ
.ÿtempfileÿdataset

.ÿquietlyÿsaveÿ`dataset'

.ÿ
.ÿ//ÿSetÿup
.ÿquietlyÿkeepÿifÿeventÿ!=ÿ"0"

.ÿquietlyÿbysortÿcaseidÿyearÿ(episode):ÿkeepÿifÿ_nÿ==ÿ1

.ÿrenameÿeventÿevent0

.ÿgenerateÿstrÿtagÿ=ÿ"<="

.ÿ
.ÿtempfileÿmarked

.ÿquietlyÿsaveÿ`marked'

.ÿ
.ÿ//ÿIfÿyouÿjustÿwantÿtoÿtagÿtheÿfirstÿnonzeroÿeventÿperÿcaseidÿandÿyear
.ÿmergeÿ1:1ÿcaseidÿyearÿepisodeÿusingÿ`dataset',ÿassert(matchÿusing)ÿnogenerateÿnoreport

.ÿ
.ÿsortÿcaseidÿyearÿepisode

.ÿlistÿcaseidÿyearÿepisodeÿeventÿtag,ÿnoobsÿsepby(caseidÿyear)ÿabbreviate(20)

ÿÿ+----------------------------------------+
ÿÿ|ÿÿcaseidÿÿÿyearÿÿÿepisodeÿÿÿeventÿÿÿtagÿ|
ÿÿ|----------------------------------------|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ1ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿ0ÿÿÿÿÿÿÿ|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ1ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿBÿÿÿÿ<=ÿ|
ÿÿ|----------------------------------------|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ2ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿBÿÿÿÿ<=ÿ|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ2ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿ0ÿÿÿÿÿÿÿ|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ2ÿÿÿÿÿÿÿÿÿ3ÿÿÿÿÿÿÿBÿÿÿÿÿÿÿ|
ÿÿ+----------------------------------------+

.ÿ
.ÿ//ÿIfÿyouÿwantÿtoÿprovideÿtheÿfirstÿnonzeroÿeventÿvalueÿ.ÿ.ÿ.
.ÿuseÿ`marked',ÿclear

.ÿdropÿepisode

.ÿmergeÿ1:mÿcaseidÿyearÿusingÿ`dataset',ÿassert(matchÿusing)ÿnogenerateÿnoreport

.ÿ
.ÿsortÿcaseidÿyearÿepisode

.ÿlistÿcaseidÿyearÿepisodeÿeventÿevent0,ÿnoobsÿsepby(caseidÿyear)ÿabbreviate(20)

ÿÿ+-------------------------------------------+
ÿÿ|ÿÿcaseidÿÿÿyearÿÿÿepisodeÿÿÿeventÿÿÿevent0ÿ|
ÿÿ|-------------------------------------------|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ1ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿ0ÿÿÿÿÿÿÿÿBÿ|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ1ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿBÿÿÿÿÿÿÿÿBÿ|
ÿÿ|-------------------------------------------|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ2ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿÿBÿÿÿÿÿÿÿÿBÿ|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ2ÿÿÿÿÿÿÿÿÿ2ÿÿÿÿÿÿÿ0ÿÿÿÿÿÿÿÿBÿ|
ÿÿ|ÿ1100122ÿÿÿÿÿÿ2ÿÿÿÿÿÿÿÿÿ3ÿÿÿÿÿÿÿBÿÿÿÿÿÿÿÿBÿ|
ÿÿ+-------------------------------------------+

.ÿ
.ÿexit

endÿofÿdo-file


.

I don't know why you think you need to loop.

[Nick Cox]

Please use dataex as requested in the FAQ Advice.

Also, please follow the request to explain where community-contributed programs you refer to come from. anymatch [not "anywatch"] is a program by Roy Wada on SSC. I've never used it, so can't comment.

More importantly, I can't work out what you want with confidence, but this may help.

The first date or time (subject to conditions) is just the minimum date or time (subject to those conditions).

More at Stata FAQ https://www.stata.com/support/faqs/d...t-occurrences/

clear 
input caseid    year    episode    str1 event
1100122    1    1    0
1100122    1    2    B
1100122    2    1    B
1100122    2    2    0
1100122    2    3    B
end 

sort caseid year episode 
by caseid: gen counter = _n 
by caseid year: egen when_first = min(cond(event != "0", counter, .)) 
gen was_first = counter == when_first 

list, sepby(caseid year) 

     +------------------------------------------------------------------+
     |  caseid   year   episode   event   counter   when_f~t   was_fi~t |
     |------------------------------------------------------------------|
  1. | 1100122      1         1       0         1          2          0 |
  2. | 1100122      1         2       B         2          2          1 |
     |------------------------------------------------------------------|
  3. | 1100122      2         1       B         3          3          1 |
  4. | 1100122      2         2       0         4          3          0 |
  5. | 1100122      2         3       B         5          3          0 |
     +------------------------------------------------------------------+

[Ankita Shukla]

Thanks Joseph. Thanks for the several hints. It was very helpful.