SAS中文论坛

标题: 求助！ [打印本页]

作者: liusiyu195 时间: 2015-4-12 11:02
标题: 求助！
下方为一个宏程序，作用是模拟一系列二分变量的LCA数据，那么如何改成模拟多级计分（3,5,7级计分）的程序呢？

%MACRO SimulateLcaDataset(  true_gamma_dataset=,
                                                      true_rho_dataset=,
                                                      output_dataset_name=,
                                                      total_n= ); /*
      PROC IML;
            * Process input...       ;
            USE &true_gamma_dataset;
                     READ ALL INTO true_gamma;
            CLOSE &true_gamma_dataset;
            USE &true_rho_dataset;
                     READ ALL INTO true_rho_yes;
            CLOSE &true_rho_dataset;
            total_n = &total_n;
         true_nclass = NCOL(true_rho_yes);
            n_items = NROW(true_rho_yes);
            ***************************************************** ;
            items = ('Item' + CHAR((1:n_items),3));
                                                                              * assumes < 1000 items;
            CALL CHANGE(items," ","0",0);
            true_class = J(&total_n,1,0);
            DO this_person = 1 TO &total_n ;
                     temp = 0;
                     CALL RANDGEN(temp, 'TABLE', true_gamma);
                     true_class[this_person] = temp;
            END;
            responses = J(&total_n, n_items, 0);
            DO this_person = 1 TO &total_n ;
                     DO this_item = 1 TO n_items;
                              temp = 0;
                              this_persons_class = true_class[this_person];
                              CALL RANDGEN(temp, 'BERNOULLI',
                                             true_rho_yes[this_item,this_persons_class]);
                     responses[this_person,this_item] = 2 - temp;
                     END;
            END;
            ResponsePatterns = J(2**n_items, n_items, 0);
            DO this_item = 1 TO n_items;
                     temp = REPEAT(1, 2**(n_items-this_item), 1) // REPEAT(2, 2**(n_items-this_item), 1);
                     temp = REPEAT(temp, 2**(this_item-1),1);
                     ResponsePatterns[,this_item] = temp;
            END;
            Count = J((2**n_items),1,0);
            DO pattern_index = 1 TO (2**n_items);
                     Matches = J(&total_n, 1, 1);
                     DO this_item = 1 TO n_items;
                              Matches =       Matches #
                                                      (responses[,this_item]=ResponsePatterns[pattern_index,this_item]);
                              Count[pattern_index] = Matches[+];
                     END;
            END;
               responses_matrix =  ResponsePatterns || Count;
            cols = items || 'Count';
            CREATE &output_dataset_name FROM responses_matrix [ COLNAME = cols ];
                     APPEND  FROM responses_matrix;
                              * The name of the append command can be misleading. ;
                              * It does not append to the data from the previous ;
                              * replication, which is replaced.  It just appends ;
                              * to the new empty dataset, i.e., writes the data. ;
            CLOSE &output_dataset_name;
      QUIT;
      DATA &output_dataset_name;
            SET &output_dataset_name;
            WHERE Count > 0; * omit non-occurring patterns;
      RUN;
%MEND;

欢迎光临 SAS中文论坛 (https://mysas.net/forum/)