Non-numeric string variable with missing value to numeric variable that recognises the missing value

[Keshab Parajuli]

I have a string variable in the form as below:

delivery

air
ship
ship
air
air
NA
NA
ship
air
air

I am trying to convert the above strings with non-numeric values such that air is assigned a label value of 0 and ship is assigned a label value of 1, while also recognising NA as missing value.

So far I have tried destring/generate, resulting in the following error:

[destring delivery, ignore ("NA") gen (delivery_new)]

[delivery: contains characters not specified in ignore(); no generate]

Similarly, I tried encode, which converted strings to numeric successfully, but still does not recognise the missing values:

[label define deliverylabel 0 "ship" 1 "air"
encode delivery, gen (delivery_new) label (deliverylabel)]

I am not sure what is the best way to tackle this. Any insights will be hugely appreciated.

Thanks,
Keshab

[Joseph Coveney]

Try something like the following

label define Modes 0 air 1 ship .n NA
encode delivery, generate(delivery_new) label(Modes)

[Keshab Parajuli]

@Joseph Coveney: Thanks much; that did work quite nicely after replacing .n with "."

I am loving this forum as it is making the start of my stata journey much more effective.

[daniel klein]

that did work quite nicely after replacing .n with "."

Thanks for reporting back. Your statement seems a bit strange, though; you cannot define a label for system missing values. Thus,

label define Modes 0 air 1 ship . NA

will exit with the error message

may not label .

You probably meant that you followed the encode command with something like

replace delivery_new = . if delivery_new == .n

which is usually not necessary because extended missing values are treated the same way as system missing values (one exception is multiple imputation, where extended missing values are treated differently).

Best
Daniel