Replace using patterns

[DF Moreno]

Hello everyone,
I am dealing with a dataset that has variable names with the same prefix. I would like to move the contents of one variable to another using that prefix as a pattern.
Let's say I have the following varnames:

var1_x var2_x var3_x var1_y var2_y var3_y

I would like to move the contents from *_x to *_y.

I thought using a foreach or replace could solve the problem, like the following: replace *_y = *_x
But it does not work. Sure I am doing something wrong. Do you know if there is a better way?

Thank you so much!


David

[daniel klein]

Why not simply drop the *_y variables and rename the *_x variables?

drop *_y
rename *_x *_y

If this is not what you want, we might need more details from your side.

[DF Moreno]

I am sorry I have not explained more details.
This replace is only for specific observations (rows), and only for some variable names (not all of them share prefixes).
So I was thinking about doing:

replace *_y = *_x if id=="some_row"

Thanks

[daniel klein]

This replace is [...] only for some variable names (not all of them share prefixes).

This makes things more complicated. You need to get a list of prefixes that you want, then implement a loop:

local list_of_prefixes ...

foreach prefix of local list_of_prefixes {
    
    replace `prefix'_y = `prefix'_x if exp
    
}

You might be able to get the list of prefixes from

unab list_of_prefixes : *_y

but I cannot be sure.

Also, your if condition, despite being pseudo-code, looks rather ad-hoc. It is usually better to identify observations by conditions, e.g., if missing(prefix_x), than referring to some ID variable.

[DF Moreno]

That worked perfectly!

Thanks