The grouping variable is given by the GROUPING parameter: GROUPING=<gr.vble1> or it can be pointed by the mask G. The grouping structure of the grouping variable may be given in the form <gr.vble>=value1(label1),value2(label2), ... The labels given in parenthesis are optional. If the grouping values are given in ascending order then they are considered as cutpoints for forming groups, e.g. KNOWLDGE=20,50,80,110,140 Values greater than the last cutpoint are treated as missing values. The list above could be shortened by the notation 20(30)140. You can use string variables as grouping variables. The grouping structure of a string variable is given in the form <gr.vble>=symbol1,symbol2, ... Note that only the first 8 characters of a string variable are significant. If the grouping structure of the grouping variable is not given then all existing values of that variable are used. The maximum number of groups is 32 (can be altered by the parameter setting MAXNCL=<new limit>). The capacity of the program is mainly limited by the requirement g*(p+1)*p/2 < 64000 where q is the number groups and p is the number of variables used for forming the classification functions. A = More information on the DISCR operation