dm log 'clear'; libname hosp 'D:\sas'; data labs; input id prelab1 prelab2 postlab1 postlab2; datalines; 1 9.8 318 10.9 215 2 . . . 129 3 . . 5.7 136 4 7.8 124 . . 5 . . 6.6 123 6 . . . 127 7 . . 5.8 127 8 7.9 196 . . 9 . . 4.9 134 10 . . . 131 11 . 171 . . 12 7.1 140 7.1 88 13 . 160 . 135 14 6.9 107 6.5 154 15 . . 6.6 122 16 . 127 . . 17 6.5 102 . . 18 . 149 . . 19 . . 6.6 120 20 . . . 129 21 . 128 . . 22 5.5 146 5.3 136 23 . . 6.5 . 24 6.6 127 6.3 130 25 . 155 . 133 26 7 145 . . 27 . . 8.1 148 28 6.6 103 . . 29 6.8 140 6.5 124 30 . 143 . . 31 . . 5.7 128 32 6.6 156 7.2 . 33 . . . 140 34 . . . 134 35 . . 6.5 111 36 . . . 137 37 5.7 127 . . 38 . 148 6.3 142 39 . 173 6.8 150 40 5.7 163 . . 41 . 150 . . 42 . 167 . . 43 8.6 184 6.5 . 44 8.2 113 7.8 173 45 24 . 144 . . 46 . . . 126 47 6.5 109 6.9 115 48 7.7 184 7.8 190 49 . 158 . . 50 . . . 127 51 . 146 . . 52 . . 5.9 128 53 . 127 5.9 131 54 9.6 198 7 84 55 . . 7 93 56 6 133 . . 57 6.7 185 7.5 204 58 6.3 165 . 136 59 4.9 150 . 137 60 6.9 130 6.9 127 61 6.3 137 7.1 166 62 . 138 . 149 63 5.5 160 5.7 152 64 12.5 592 5.5 147 65 . . . 148 66 . . . 161 67 7.4 162 6.5 204 68 9 255 5.8 150 69 7.3 141 7.2 128 70 7.5 . 7.7 198 71 6.9 173 . . 72 . . 6.9 150 73 6.3 135 6.3 164 74 8.5 110 8.3 156 75 6.7 121 7 125 76 . . 5.7 132 77 . . . 154 78 9.6 201 8 . 79 . 138 . . 80 . 131 . . 81 . 183 . 141 82 . . . 185 83 6.3 145 . . 84 . 131 . 134 85 . 185 6.5 122 86 . . . 201 87 7.8 160 8.3 213 88 5.6 138 . . 89 10.9 114 8.9 178 90 . . 6.6 105 91 . . . 147 92 6.7 103 6 138 93 . . . 139 94 . 137 . . 95 . . . 134 96 . . 6.2 161 97 9 191 7.3 156 98 6.7 107 . . 99 . . 9.1 127 100 . . 6.6 119 ; run; proc print; run; /*I have a dataset of the lab values of two tests before and after taking a certain drug. Using a single macro, I need to do three things: 1. remove missing values for the changes in these lab values (post-pre). That is changelab1 and changelab2 2. classify the patients in this sample into three groups: Group A: Those with prelab1 greater than or equal 6 Group B: Those with prelab2 greater than or equal 130 Group C: Those with prelab1 >= 6 and prelab2 >= 130 3. do proc univariate on each of each of the three groups to find the number of patients who have done each lab test, and the mean and standard deviation of each lab test for each group. So, I need to get finally: The number of people who have done lab1 in groups A, B, and C (pre and post). Supposably they are equal. The mean value of lab1 and lab2 values (pre and post) in the three groups. The standard deviation of lab1 and lab2 values (pre and post) in the three groups.   Is that possible using a single macro? ... View more

I have a dataset that I need help with.   data help; input modelstep var1 $ var2 $ var3 $ var4 $ var5 $ outcome auc; datalines; 0 v1 v2 v3 v4 v5 1 0.003 1 v1 v2 v3 '' v5 1 0.004 2 v1 v2 v3 '' '' 1 0.007 3 '' v2 v3 '' '' 1 0.01 4 '' '' v3 '' '' 1 0.02 0 v1 v2 '' v4 v5 2 0.005 1 v1 '' '' v4 v5 2 0.006 2 v1 '' '' '' v5 2 0.02 3 v1 '' '' '' '' 2 0.03 ; run; This dataset represents an iterative process of backward selection of model variables where I am required to identify (for each outcome) the model step at which a specified number of variables yields an AUC difference of 0.01, and then specify the previous model step variables as my model of choice. I want to do that using a lag function to get the table outlined below. Because every row represents a model step (that is a number of variables leading to the desired AUC difference, I would like to specify the row just before the model leading to the AUC difference of 0.01 for each outcome.*/ Any help with that, please? Thanks ... View more