topic Creating a Unique Identifier using Name, DOB, and Gender in New SAS User

Creating a Unique Identifier using Name, DOB, and Gender

suzannep — Wed, 12 Jun 2019 15:00:20 GMT

Hello,

I am trying to create a unique identifier using parts of first and last name, DOB and Gender. I want the first 2 letters of the first_name & the first 2 letters of the last_name & last 2 letters of the last_name & DOB with no slashes & Gender as either M or F.

For example:

First_name Last_name DOB Gender ID_number

John Smith 01/02/1995 M JOSMTH01021995M

Mary Jackson 03/15/1975 F MAJAON03151975F

Any help would be greatly appreciated!!

Re: Creating a Unique Identifier using Name, DOB, and Gender

PaigeMiller — Wed, 12 Jun 2019 15:56:07 GMT

Assuming DOB is an actual numeric SAS date value and not a character string

data want;
    set have;
    id = cats(substr(first_name,1,2),substr(last_name,1,2),
        reverse(substr(reverse(trim(last_name)),1,2)),put(dob,mmddyyn8.),gender);
run;

If the last name has less than 4 letters, such as the name "Ty Law", then your ID would begin with TyLaaw (the letter A being repeated by this algorithm)

Re: Creating a Unique Identifier using Name, DOB, and Gender

suzannep — Wed, 12 Jun 2019 18:24:51 GMT

Thank you! my date is a character string. What is the format to use so that there are no / in the id?

Re: Creating a Unique Identifier using Name, DOB, and Gender

PaigeMiller — Wed, 12 Jun 2019 18:28:37 GMT

data want;
    set have;
    id = cats(substr(first_name,1,2),substr(last_name,1,2),
        reverse(substr(reverse(trim(last_name)),1,2)),compress(dob,'/'),gender);
run;

Re: Creating a Unique Identifier using Name, DOB, and Gender

suzannep — Wed, 12 Jun 2019 18:37:05 GMT

THANK YOU!!!

Re: Creating a Unique Identifier using Name, DOB, and Gender

ballardw — Wed, 12 Jun 2019 20:16:43 GMT

I would NOT call this a "unique id". In one data set I work with of about 5000 individuals I had two people with the same first and last name, birth date and gender. I knew they were not the same person from other data such as residence.

Since Joel Smutny would get the same letters as John Smith, your algorithm is very likely to have issues with the "uniqueness" if the number of people involved is not even very large.