How to convert weekly to yearly

[hengxu]

I have a question about this:

I have weekly data, but that is described in number of weeks, i.e., (1,2,3,....,400). The first week is the week of 1990/09/01, and so on so forth. My goal is to convert these weeks into year, for example, a new variable "year" likes (1990, 1991,....)

Thanks so much for your help!

[ben earnhart]

Well, seems actually easier to go straight to year than to convert to dates.

clear
set obs 400
gen theweek=_n
gen year=1990+floor((theweek-1)/52)

[hengxu]

Well, seems actually easier to go straight to year than to convert to dates.

clear
set obs 400
gen theweek=_n
gen year=1990+floor((theweek-1)/52)

Thank you Ben

Another problem is my data is panel and for each week I also have many variables (e.g., many stores sales information), so it is not easy to simply gen obs.

Could you please help on converting dates?

[ben earnhart]

please paste a chunk of your data (and read the FAQs on how to do so). I assumed you were merely concerned with the single variable that was the week #. In my code, you could ignore the first three lines -- I was just generating observations 1-400 to try to simulate your data. Is the data already in date format, and is it long or wide?

[Attaullah Shah]

This might be of help http://www.opendoors.pk/STATA/cumula...o-monthly-data

[Nick Cox]

Weeks outside Stata don't correspond to weeks inside Stata.

So, avoid week functions.

If you convert to daily dates

gen day = mdy(9,1,1990) + (week - 1) * 7

then you can extract the year using year():

gen year = year(day)

See (e.g.) http://www.stata-journal.com/article...article=dm0052 and later notes by the same author on weeks in the Stata Journal.

(Having panel data or other variables seems of no consequence here.)

[hengxu]

Weeks outside Stata don't correspond to weeks inside Stata.

So, avoid week functions.

If you convert to daily dates

gen day = mdy(9,1,1990) + (week - 1) * 7

then you can extract the year using year():

gen year = year(day)

See (e.g.) http://www.stata-journal.com/article...article=dm0052 and later notes by the same author on weeks in the Stata Journal.

(Having panel data or other variables seems of no consequence here.)

Thank you Nick, That works.

Regards.