Trying to understand formats

[bhca60]

Hi, I just wanted to see if someone could explain what this means exactly:

data disp;
     set trcb.DispositionCode_Xref;

     length start $2 label $100;
     fmtname = '$disposition';

     start = Disp_Code;
     label = Disp_Code_Desc;

     keep start label fmtname;
run;

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

I am not well-versed on formats that much. I get the gist but when I look at this I'm a bit lost on what exactly this means. Thank you, any help is appreciated.

[ballardw]

It appears that the data set is intended to be used with the CNTLIN option of Proc Format to create a format named DISPOSITION for use with character values.

The structure of a CNTLIN data set requires certain variables, minimum is the name of the format which must be in a variable named FMTNAME, Start as the value and Label as the text to display instead of the value of Start.

There are other variables to indicate other properties set as options.

If you know how to write small custom format you could use the CNTLOUT= option on proc format to create a data set that describes the format(s). That may help understand a bit better what the variables must look like.

The Sort is not needed and could actually be improper if the set had more than one value of FMTNAME as the Format procedure would expect all of the lines for a single format to be together in the data set. The NODUPEKEY is there to prevent errors of multiple observations with the same Start value. Again proc format except for very specific purposes expects to see values in the Start (or End if a range of numeric values is used) only once.

I would expect to see something like this to actually create the format for use:

Proc format <library=somelib> cntlin=disp;
run;

 This would be similar to writing a Proc format value statement like this:

Proc format;
Value $disposition
'01' = 'Meaning for code 01'
'02' = 'Meaning for code 02'
;
run;

The '01' '02' are the START values. The 'Meaning for code' text are the values of LABEL and $disposition is the FMTNAME value.

[bhca60]

Thank you!

[data_null__]

You have code to create a PROC FORMAT CNLTLIN= data set.

To create the character value label format $DISPOSITION using this data you need to this.

proc format CNTLIN=disp;
   quit;

The you would be able to use $disposition. in a format statement or PUT function.

[Tom]

What part do you not understand?

The SAS code you shared?

The DATA step is making a new dataset from an old dataset.  The dataset it is making can be used to create a FORMAT.

The PROC SORT step is to order the observations and make sure the same START value does not appear more than once.

How to convert that dataset into a format?

Use PROC FORMAT with the CNTLIN= option.

proc format cntlin=disp;
run;

What a FORMAT does?

Formats are used to convert values into text.  In this case the values are character strings up to two bytes long.  Your dataset can be used to make a format named $DISPOSITION that converts values that match the values of DISP_CODE into text strings of the corresponding value of DISP_CODE_DESC.

How to use a FORMAT?

You would use the $DISPOSTION format with any variable that has the same set of 2 letter codes that was in your dataset so that they are printed/displayed/converted to text.  Say you have a dataset named EVENTS that had a variable named EVENT_DISP that was using those same 2 letter codes.  You could then get a summary of how many events ended with each disposition code using a step like this:

proc freq data=events;
  tables event_disp;
  format event_disp $disposition. ;
run;