generating a new variable based on row numbers - issues with 'gen time = _n'

[Kishan Bakrania]

Hi,

I am trying to run the command 'gen time = _n' in order to create a new variable 'time' - however my dataset is very large ~ 100,000,000 observations with 3 variables. I need to create this 'time' variable in order for the next step of my analysis. The issue I am having after running this command is as follows;

At some points in this newly generated 'time' variable, I get repeated values - for example instead of observing;

.
.
.
56030
56031
56032
.
.
.

I see the following;

.
.
.
56030
56030
56031
.
.
.


etc.

this happens at several different points over the dataset and I can't continue with the next stage of my analyses (Since I need to use 'tsset time' and get the error r(451) - repeated time values in sample)

Does anyone know how I can solve this?

many thanks,

Kishan

[Richard Williams]

Try

gen long time = _n

Do you really have 100 million different times though? I wonder if maybe you instead need something like

bysort id: gen time = _n

But even that may not be quite right. Maybe you could describe your data a bit more, including the information that identifies cases. I am surprised there isn't some sort of time variable already. What keeps the records properly sorted by time?

[Kishan Bakrania]

Thank you Richard,

That was such a simple solution!

gen long time=_n worked well.

I have accelerometer data which is measured over 10 days at 100 hz which results in the large dataset.

Thank you very much.