How to Use PROC FORMAT in SAS (With Examples)

You can use PROC FORMAT in SAS to create a mapping of data values into data labels.

This procedure uses the following basic syntax:

proc format;
    value points_range
        25-high='High'
        15-<25='Medium'
        other ='Low';
run;

This particular example creates the following mapping:

Values equal to 25 or greater will be shown as ‘High‘
Values between 15 and 25 will be shown as ‘Medium‘
All other values will be shown as ‘Low‘

The following examples show how to use PROC FORMAT with the following dataset in SAS:

/*create dataset*/
data my_data;
    input team $ position $ points;
    datalines;
A Guard 25
A Guard 20
A Guard 30
A Forward 25
A Forward 10
B Guard 10
B Guard 22
B Forward 30
B Forward 10
B Forward 10
B Forward 25
;
run;

/*view dataset*/
proc print data=my_data;

Example 1: Use PROC FORMAT to Format Values as Labels in Frequency Table

Suppose we use PROC FREQ to create a frequency table of values in the points column of the dataset:

/*calculate frequency of values in points column*/
proc freq data = my_data;
    table points;
run;

The output displays the frequency of each individual value in the points column.

However, suppose we would like to format the values as follows:

Values equal to 25 or greater will be shown as ‘High‘
Values between 15 and 25 will be shown as ‘Medium‘
All other values will be shown as ‘Low‘

We can use PROC FORMAT to do so:

/*define formatting for points variable*/
proc format;
    value points_range
        25-high='High'
        15-<25='Medium'
        other ='Low';
run;

/*create frequency table for points variable, using formatting defined above*/
proc freq data = my_data;
    table points;
    format points points_range.;
run;

The frequency table now groups the values of the points variable into the labels that we specified using the PROC FORMAT statement.

Example 2: USE PROC FORMAT to Create New Variable

We can also use PROC FORMAT to create a new variable in a dataset that converts data values into data labels.

The following syntax shows how to do so:

/*define formatting for points variable*/
proc format; 
    value points_range
        25-high='High'
        15-<25='Medium' 
        other ='Low';
run;

/*create new dataset with points_range variable*/
data new_data;
    set my_data;
    point_range = put(points, points_range.);
run;

/*view dataset*/
proc print data=new_data;