Before processing the large data set, it is a good practice to check the levels of the categorical varaible and make sure records are assigned to the right category and missing categories are treated.
Use _character_ in the freqency procedure to apply the procedure to all character variables. Be careful not to include any character variables that are numeric in nature, eg id variable in the data set. Using “nlevels” option to provide a summary table of number of the levels, missing levels, non-missing levels for each categorical variable. Then check the frequency table for the specific categorical variable for the values of the levels.
proc freq data= table1 nlevels; tables _character_ / nocum nopercent out=freqcnt; run;
| Number of Variable Levels | ||||
|---|---|---|---|---|
| Variable | Label | Levels | Missing Levels | Nonmissing Levels |
| gender | Gender Code | 4 | 0 | 4 |
| Var2 | label2 | 12 | 1 | 11 |
| Gender Code | |
|---|---|
| gender | Frequency |
| F | 495320 |
| M | 402399 |
| U | 3475 |
| {NUL} | 1 |