topic Re: NA are called unknown in the dataset in SAS Software for Learning Community

NA are called unknown in the dataset

smartis — Thu, 20 Jan 2022 10:38:40 GMT

Goodmorning everyone!

I uploaded a dataset on Sas Viya for Learners.

My problem is: the NA values are called unknown.. how can I recall them with the classic name: 'NA' so that SAS recognise and process them as NA?

Thank you so much!

Martina

Re: NA are called unknown in the dataset

Kurt_Bremser — Thu, 20 Jan 2022 11:42:42 GMT

I guess(!) that you actually did not upload a dataset, but a data file that was then imported into SAS and has a numeric column with 'NA' in some places. Is that correct?

Re: NA are called unknown in the dataset

smartis — Thu, 20 Jan 2022 13:23:52 GMT

not really, sorry, probably I didn't explain well myself:

In the dataset I'm working on there are, in different columns, values called 'unknown', when I perform the imputation, SAS tell me that there aren't missing values, but actually there are and they are the unknowns!

So know: how can I substitute that 'unknown' into NA's??

Thanks in advance

Martina

Re: NA are called unknown in the dataset

Kurt_Bremser — Thu, 20 Jan 2022 16:03:59 GMT

OK, so we need to know

the type of the source variable
the format assigned to it
if character, the defined length will also be good to know
the raw content (displayed without a format)

Run a PROC CONTENTS on the dataset. If you find a format attached, run this:

data _null_;
set dataset (obs=10);
format var; /* removes any format */
put var;
run;

(replace "dataset" and "var" with your dataset and variable name, respectively)

This will write the first 10 values to the SAS log, from where you can copy/paste them here.

Re: NA are called unknown in the dataset

Reeza — Thu, 20 Jan 2022 16:27:44 GMT

NAs are a concept in R/Python.

Nulls/Missing are a DB concept

SAS uses spaces/blanks for character missing and . (period) for numeric missing.

Is this a correct interpretation of your question:

You have a data set that you're trying to impute missing data
The field is character
One of the values in the field is 'unknown' and SAS is not treating the unknown as missing
You want to impute for the 'unknown values in the column

If this is correct, you first need to recode the unknown to missing

data missing;
set have;

if varInterest='unknown' then call missing(varInterest);

run;

@smartis wrote:

not really, sorry, probably I didn't explain well myself:

In the dataset I'm working on there are, in different columns, values called 'unknown', when I perform the imputation, SAS tell me that there aren't missing values, but actually there are and they are the unknowns!

So know: how can I substitute that 'unknown' into NA's??

Thanks in advance

Martina

Re: NA are called unknown in the dataset

smartis — Sun, 23 Jan 2022 07:02:01 GMT

Thank you so much and sorry for my late answer!

Martina