generating new variable that sums many variables

[LarsFolkestad]

Dear Statalisters

I have 25 variables all named event_eventtype , where event type may differ.
I then have an additional 25 variables named event_eventtype180, event type can be both_sides or oneside - the naming seems a bit random.
I then have an additional 25 variables named event_eventtype365

all variables are integer 0,1,2,3,4…_N


i want to generate a new variable that is the sum of all 25 event_eventtype variables
and one that is the sum of all event_eventtype180
and one that is the sum of all event_eventtype365

the slow and space filling way is to copy all variable names of interest into the do editor and enter + between the variables, but i imagine there is a better solution to my problem.


Hope this makes sens

thank you

lars

[Stephen Jenkins]

Does the following pseudo-code help? (1) Use ds to get a list of all your "event_eventtype" variables. (2) note that ds leaves behind the varlist in r(varlist). (3) Feed this macro containing the varlist to egen: egen newvar = rowtotal(`r(varlist)')

[Nick Cox]

ds is a wonderful command. But it may be that

 
describe *side*180 
describe *side*365

shows that a wildcard will work in each case.

[LarsFolkestad]

but is there a way to leave out variables that ends in 180 and 365 from the ds r(varlist)?

Nick,
I seemed to have left out that the _eventname is random i only selected 2 to show that there were different names, it can be up, down, right, left and all other directions really.

lars

[Nick Cox]

ds has a not option.

You will thus need more than one wildcard.

[LarsFolkestad]

yes, but how do i use the option?
My variables are all named
event_aneventname
event_aneventname180
event_aneventname365

and the not option does not seem to allow variable names as i get an error message saying: event_ option not allowed when i go:

ds event_*, not event_*180 event_*365

and further more, I'm not sure how to do the rowtotal when i only want the *180's or *365's because i cannot see how i leave out the event_aneventname variables.

Lars

[Brendan Cox]

ds, not is a bit counter-intuitive (in my opinion). If your variables are frog toad newt cat caterpillar, and you want to return frog toad newt, you can use ds cat*, not. ds cat* will return cat and caterpillar, and the not option will 'invert' the varlist to return the remaining variables. You don't actually specify any variables with/after the not option.

[Nick Cox]

Another technique is

unab all: event_* 
unab A : event_*365 
unab B : event_*180 
local some : list all - A 
local some : list some - B 
di "`some'"

[Phil Bromiley]

There might be an easier way to do this with egen and rowtotal with wildcards. Something like

egen a=rowtotal(event_*)
egen a=rowtotal(event_*180)

[LarsFolkestad]

But Phil,
egen a=rowtotal(event_*), will include all variables both event_*, event_*180 AND event_*365,
I am interested in generating
all= sum of all event_* (not including event_*180 or event_*365)
180= sum of all event_*180 (not including all or event_*365)
365= sum of all event_*365 (not including all or 180)

I have a slight log on issue this morning so i cannot test the above given examples.

Thank you anyway.

Lars

[Nick Cox]

I think you have all the techniques you need.

[LarsFolkestad]

It worked out fine, thank you.