topic Re: Convert continuous variable into categorical in SAS Programming

Convert continuous variable into categorical

Rodrigue — Thu, 15 Jul 2021 11:47:11 GMT

Hello,

I have a continuous variable (age) and I would like to create a new variable (age2) which is a categorical (0-20; 21-40;41-60;61-80;81-100). I used the format but it doesn't work. Could you help me ?

Thanks

Re: Convert continuous variable into categorical

PaigeMiller — Thu, 15 Jul 2021 12:04:38 GMT

The first error message gives you the answer. You cannot have a format name that ends in a number. So AGE2 cannot be a format name, but AGETWO is fine.

May I suggest that instead of "1", "2", etc., you use more meaningful formats, such as

value agetwo 0-20="0-20" 21-40="21-40" ... ;

Also note that there is no semi-colon between the format levels.

Re: Convert continuous variable into categorical

Rodrigue — Thu, 15 Jul 2021 12:10:10 GMT

Thank you.

I have this now. Do you know what is wrong please ?

Re: Convert continuous variable into categorical

PaigeMiller — Thu, 15 Jul 2021 12:19:29 GMT

Please read the last sentence of my earlier reply.

Re: Convert continuous variable into categorical

Rodrigue — Thu, 15 Jul 2021 12:23:24 GMT

Thank you. Sorry for not reading completely.
Now the program runs perfectly but I don't have the new variable in "try". How I can add the new column "agecat" to the dataset ?

Re: Convert continuous variable into categorical

PaigeMiller — Thu, 15 Jul 2021 12:25:35 GMT

Formats do not add a new variable. They change the way the variable appears to us humans, and they change the way most SAS PROCs work with the variable. So most SAS PROCs will use your continuous variable as if it was categories.

Re: Convert continuous variable into categorical

Rodrigue — Thu, 15 Jul 2021 12:35:37 GMT

Thank you.

I tried this code but I only have . in the "agecat" column. Any thoughts ?

Re: Convert continuous variable into categorical

Tom — Thu, 15 Jul 2021 12:35:53 GMT

A format specification needs to include the period otherwise the SAS compiler will see that name as the name of a variable instead of as a reference to a format.

format age agecat. ;

The syntax you currently have does not include a format specification:

format age agecat ;

So it will remove any formats that were already attached to the AGE and AGECAT variables. If the variable AGECAT does not exist then it will be created as numeric and will have missing values on all observations.

Re: Convert continuous variable into categorical

PaigeMiller — Thu, 15 Jul 2021 12:41:20 GMT

@Rodrigue wrote:

Thank you.

I tried this code but I only have . in the "agecat" column. Any thoughts ?

Important concept: READ THE LOG.

And if you can't really understand what it is telling you, then SHOW US the log. COpy and paste the log as text (not as a screen capture) into the window that appears when you click on the </> icon. (No more screen captures, we want text, posted in the appropriate code or text box)

Re: Convert continuous variable into categorical

Rodrigue — Thu, 15 Jul 2021 12:48:40 GMT

I can't copy paste the log but only do screen shots. Sorry. I understand it is a problem of character/numeric data. Isn't it ?

Re: Convert continuous variable into categorical

PaigeMiller — Thu, 15 Jul 2021 13:00:47 GMT

Do the notes in the log give you a clue as to what the problem is?

Why have you switched from using FORMATs to this method?

Re: Convert continuous variable into categorical

Rick_SAS — Thu, 15 Jul 2021 14:05:53 GMT

You initialize AGECAT=., which defines the variable as a numeric variable. But then you try to assign AGECAT="0-19", which is a character string.

Re: Convert continuous variable into categorical

Rodrigue — Thu, 15 Jul 2021 14:35:47 GMT

Thank you Rick