topic character variables to numerical variables in SAS Programming

character variables to numerical variables

archibald — Mon, 15 Aug 2016 15:27:18 GMT

I have a character variable representing various US states that I need to convert to numberical.

data have;

State

data want;

State_num

so one so forth

I know I can achieve this using :

If state=”Kentucky” then state_num=1;

If state=”Louisiana” then state_num=2;

and so one so forth.

But this would be lenghty, does anyone know what would be the optimal way of converting from character to numeric.

Thanks.

Re: character variables to numerical variables

KachiM — Mon, 15 Aug 2016 15:33:01 GMT

You can create a format that gives a unique number when the Statename is given.

Other easy way is to use _N_, the obsevation number, to be attached to each state.

The former is better.

Re: character variables to numerical variables

Reeza — Mon, 15 Aug 2016 18:03:09 GMT

There are also the FIPS code that you can use. Look into the functions. I know you can go from state abbreviation to code.

Re: character variables to numerical variables

archibald — Mon, 15 Aug 2016 18:18:25 GMT

@ datasp, would creating a format the same as recoding the variable as I shown in my posting?
Thanks

Re: character variables to numerical variables

archibald — Mon, 15 Aug 2016 18:20:04 GMT

Reeza, there are no FIPS Codes in the dataset. Plus the states have to be ordered based on crime level.

Re: character variables to numerical variables

Reeza — Mon, 15 Aug 2016 18:27:22 GMT

I think we're missing some information, please clarify your requirements.

I was suggesting mapping the State Names to FIP codes, I think there's a SAS function that maps this.

If order matters then please clarify your requirements.

Re: character variables to numerical variables

Tom — Mon, 15 Aug 2016 18:31:49 GMT

If your data is one record per state then you can define your own code.

data want ;
  set have ;
  state_num + 1;
run;

Re: character variables to numerical variables

KachiM — Tue, 16 Aug 2016 01:00:27 GMT

There is no need to list all Names to make the format. We can create format using the data set.

I show how it can be made.

Your interest is not known as how you will be using the crime rates. If it is known, an approprite solution can be given.

Besides the Format, Hash Objects and Array with PIBw. can be used.

data have;
input state &$15.;
datalines;
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
;
run;

data cnt;
retain fmtname 's2num' type 'I';
      set have end = last;
      start = state;
      label = _N_;
      output;
   if last then do;
      hlo = 'O';
      label = 0;
      output;
   end;
run;

proc sort data = cnt nodupkey;
by start;
run;

proc format  cntlin = cnt;
run;


We can verify the Format by cycling the original data set. Note the use of INPUT(STATE, s2num.) to get the unique number for a given State.

data want;
   set have;
   state_id = input(state, s2num.);
run;