计算置信区间

proc freq data=datain;
    by group;
    tables var / missprint nowarn binomial(level='1' cl=exact ) alpha=0.05;
    weight n/zero;（对发生的做置信区间，考虑发生数量为0的情况）
    ods output binomialcls=cl;
run;
ods output close;

datain的格式如下：
 
group（分组） Var（每组相对立的事件分类，eg：发生/没发生） N（每种情况下的数量）
1                       0                                                                              5
1                       1                                                                              6
2                       0                                                                              2
2                       1                                                                              4
3                       0                                                                              5
3                       1                                                                              3

将会得到每个组中发生的置信区间。