Recoding numeric variable to character

[Rohith Perike]

I'm need help in recoding numeric variables to characters.

[Maarten Buis]

That depends on the details of what you want to do exactly. For example, should all observations become strings (characters)? It is easiest to explain those details with an example. See the FAQ (black bar near the top of this page ) for how to do so.

[Rohith Perike]

I have a variable by name LOCAL_MONTH. I want to change the month numbers to actual month name (eg. May for 5, june for 6)

[Ken Chui]

Ah, it's better to apply a labeling scheme. Storing them as text can be inconvenient later. Here is an example (with sample data):

clear
input mos freq
5 1
6 22
7 101
8 88
9 8
end

expand freq

* Your original form
tab mos

* String format (Bad idea)
generate mos_string = "May" if mos == 5
replace  mos_string = "June" if mos == 6
replace  mos_string = "July" if mos == 7
replace  mos_string = "August" if mos == 8
replace  mos_string = "September" if mos == 9

* Tabulate: notice that months are now alphabetically ordered
tab mos_string

Result is a bit odd:

 mos_string |      Freq.     Percent        Cum.
------------+-----------------------------------
     August |         88       40.00       40.00
       July |        101       45.91       85.91
       June |         22       10.00       95.91
        May |          1        0.45       96.36
  September |          8        3.64      100.00
------------+-----------------------------------
      Total |        220      100.00

* Instead, try labeling:
label define month_name 5 "May" 6 "June" 7 "July" 8 "August" 9 "September"
label values mos month_name

* Now the order makes sense
tab mos

Here is the labeled version:

        mos |      Freq.     Percent        Cum.
------------+-----------------------------------
        May |          1        0.45        0.45
       June |         22       10.00       10.45
       July |        101       45.91       56.36
     August |         88       40.00       96.36
  September |          8        3.64      100.00
------------+-----------------------------------
      Total |        220      100.00

[Clyde Schechter]

"Friendly amendment" to #2. With only five months, I suppose writing out all five is not especially burdensome and offers the advantage of transparency. But if the full data set contains all twelve months, or a subset that is nearly that large, I think it is better to define the label in a loop:

levelsof mos, local(months)
foreach m of local months {
    label define month_name `m' "`:word `m' of `c(Months)''", add
}
label values mos month_name

[Nick Cox]

For other stored sequences, look through the results of

creturn list

and/or consult https://journals.sagepub.com/doi/pdf...867X0400400213