Question

我在SAS EG中有一张表如下：

DateTime            User    Prod    Date        Date2       User2   Prod2   DateTime2           SecDiff Calc
20MAR2014:20:17:00  54823   1430    20140320    .           .       .       .                   .       1
13MAR2014:09:07:16  66019   8244    20140313    20140320    54823   1430    20Mar14:20:17:00    -644984 1
13MAR2014:09:07:44  66019   8244    20140313    20140313    66019   8244    13Mar14:09:07:16    28      0
13MAR2014:09:08:17  66019   8244    20140313    20140313    66019   8244    13Mar14:09:07:44    33      0
13MAR2014:09:08:43  66019   8244    20140313    20140313    66019   8244    13Mar14:09:08:17    26      0
13MAR2014:09:09:12  66019   8244    20140313    20140313    66019   8244    13Mar14:09:08:43    29      0
13MAR2014:09:10:34  66019   8244    20140313    20140313    66019   8244    13Mar14:09:09:12    82      0
13MAR2014:09:11:08  66019   8244    20140313    20140313    66019   8244    13Mar14:09:10:34    34      0
13MAR2014:09:11:34  66019   8244    20140313    20140313    66019   8244    13Mar14:09:11:08    26      0
14MAR2014:21:19:18  66019   8244    20140314    20140313    66019   8244    13Mar14:09:11:34    130064  1
14MAR2014:21:19:52  66019   8244    20140314    20140314    66019   8244    14Mar14:21:19:18    34      0

所有带有2的列都是＆＃34;非-2＆＃34;列。

此数据说明了某个DateTime标记上特定产品的用户活动。

我有兴趣创建＆＃34; Sessions＆＃34;基于一些参数，但在最后一列中说明了＆＃34; Calc＆＃34;。因此，对于每个新的会话，应该有一个新的号码。参数是＆＃34; Date = Date2＆＃34;，＆＃34; User = User2＆＃34;，＆＃34; Prod = Prod2＆＃34;和＆＃34; SecDiff＆lt; = 3600＆＃34;。最终目标是计算每个会话的秒数（通过添加SecDiff）。所以在这个例子中，表格会给我这些结果：

DateTime            User    Prod    Date        Seconds  Calc
20MAR2014:20:17:00  54823   1430    20140320    .        1
13MAR2014:09:07:16  66019   8244    20140313    258      2
14MAR2014:21:19:18  66019   8244    20140314    34       3

以秒为单位返回不同的值，但会删除计算出的极值并表示新的会话。

Answer 1

我不确定我是否理解这一点，但为什么你需要所有这些列。只需使用前三个

data a;
input act_time datetime20. product user;
cards;
20MAR2014:20:17:00  54823   1430    
13MAR2014:09:07:16  66019   8244    
13MAR2014:09:07:44  66019   8244 
13MAR2014:09:08:17  66019   8244
13MAR2014:09:08:43  66019   8244
13MAR2014:09:09:12  66019   8244 
13MAR2014:09:10:34  66019   8244   
13MAR2014:09:11:08  66019   8244   
13MAR2014:09:11:34  66019   8244    
14MAR2014:21:19:18  66019   8244
14MAR2014:21:19:52  66019   8244   
;
run;

proc sort data=a out=b; by product user act_time; run;

data c(drop=last_time);
retain last_time;
set b;
by product user;
if first.user then do; 
    session + 1;
    last_time = .;
end;
else do; 
    if act_time >= last_time + 3600 then do;
       session + 1;
    end;
end;
last_time = act_time;
run;    

data d;
retain first_time;
set c;
by session;
if first.session then do;
    first_time = act_time;
end;
if last.session then do;
    duration = act_time - first_time;
    output;
end;
run;

通过SAS EG中的会话制作小组

1 个答案: