我使用proc方式计算业务线支付的份额,数据如下:
data Test;
input ID Business_Line Payment2017;
Datalines;
1 1 1000
2 1 2000
3 1 3000
4 1 4000
5 2 500
6 2 1500
7 2 3000
;
run;
我想要计算一个额外的列,按组(business_line)计算付款的百分比(重量):
data Test;
input ID Business_Line Payment2017 share;
Datalines;
1 1 1000 0.1
2 1 2000 0.2
3 1 3000 0.3
4 1 4000 0.4
5 2 500 0.1
6 2 1500 0.3
7 2 3000 0.6
;
run;
我到目前为止使用的代码:
proc means data = test noprint;
class ID;
by business_line;
var Payment2017;
output out=test2
sum = share;
weight = payment2017/share;
run;
我也试过
proc means data = test noprint;
class ID;
by business_line;
var Payment2017 /weight = payment2017;
output out=test3 ;
run;
欣赏帮助。
答案 0 :(得分:3)
Proc FREQ
将计算百分比。您可以将输出的PERCENT
列除以得到分数,或者使用下游的百分数。
在此示例中,id
与payment2017
交叉,以确保所有原始行都是输出的一部分。如果id
不存在,并且有任何重复付款金额,FREQ
将汇总付款金额。
proc freq data=have noprint;
by business_line;
table id*payment2017 / out=want all;
weight payment2017 ;
run;
答案 1 :(得分:1)
使用proc sql:
很方便proc sql;
select *, payment2017/sum(payment2017) as share from test group by business_line;
quit;
数据步骤:
data have;
do until (last.business_line);
set test;
by business_line notsorted;
total+payment2017;
end;
do until (last.business_line);
set test;
by business_line notsorted;
share=payment2017/total;
output;
end;
call missing(total);
drop total;
run;