Fill null values

[Esteban Jara]

Hello. For a data structure like the one in the figure, sorted in ascending order by "numdom" within the commune-street group, I need to fill the null values of "uv" as long as the condition that the previous value of "uv "match the next non-null. In my example, observations 4 and 5 should be filled with the value uv = "20A", but observations 9 and 10 should not, since "20B"! = "20C". How could I do it? Thank you

[Daniel Shin]

Try this (hopefully no errors):

preserve
tempfile temp_file
drop if uv==""
gen next_is_same=uv==uv[_n+]
keep comuna numdom same
save `temp_file'
restore
merge 1:1 comuna numdom using `temp_file', keep(1 3) nogen
replace same=same[_n-1] if same==. 
replace uv=uv[_n-1] if uv=="" & same[_n-1]==1

[Esteban Jara]

It did not work. It filled in values that should have remained null :-(

Thank you anyway

[Daniel Shin]

I saw an error (fixed below). Give it another go?

preserve
tempfile temp_file
drop if uv==""
gen next_is_same=uv==uv[_n+1] // FIXED
keep comuna numdom same
save `temp_file'
restore merge 1:1 comuna numdom using `temp_file', keep(1 3) nogen
replace same=same[_n-1] if same==.
replace uv=uv[_n-1] if uv=="" & same[_n-1]==1

[Esteban Jara]

I had already detected that notation error and had corrected it before ...

[Ali Atia]

I believe this will do what you want:

gen uvmiss = uv!=""
bysort comuna calle (uvmiss numdom): gen nextnonmiss = cond(_n!=_N,uv[_n+1],"N/A") if uv!=""
bysort comuna calle (numdom): replace nextnonmiss = nextnonmiss[_n-1] if nextnonmiss==""
by comuna calle: replace uv = uv[_n-1] if uv=="" & nextnonmiss==uv[_n-1]
drop nextnonmiss uvmiss