SAS：计算移动平均的方法

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

方法1：proc expand的用法
出处：http://elek.me/calculate-moving-average-using-proc-expand-in-sas.html

1. 
   
2. data have;
   
3. do group=2 to 6;
   
4.   do value=1 to 20 by 3,21,28,29,50;
   
5.    output;
   
6.   end;
   
7. end;
   
8. run;
   
9. /*1. Backward moving average: five-period
   
10. Yt = ( Xt-4 + Xt-3 + Xt-2 + Xt-1 + Xt ) / 5*/
    
11. proc expand data=have method=none out=want1(drop=time);
    
12. by group notsorted;
    
13. convert value=value1 / transform=(movave 5 trimleft 4);
    
14. run;
    
15. /*2. Forward moving average: five-period
    
16. Yt = ( Xt + Xt+1 + Xt+2 + Xt+3 + Xt+4) / 5*/
    
17. proc expand data=have method=none out=want2(drop=time);
    
18. by group notsorted;
    
19. convert value=value2 / transform=(reverse movave 5 reverse trimleft 4);
    
20. run;
    
21. /*3. Centered moving average:
    
22. 
    
23. 3.1 the centered moving time window operator is an odd number, for example, five-period
    
24. Yt = ( Xt-2 + Xt-1 + Xt + Xt+1 + Xt+2 ) / 5 */
    
25. proc expand data=have method=none out=want3(drop=time);
    
26. by group notsorted;
    
27. convert value=value3 / transform=(cmovave 5 trim 2);
    
28. run;
    
29. /*3.2 the centered moving time window operator is an even number, for example, 4-period
    
30. 3.2.1 one more lead value than lagged value is included in the time window
    
31. Yt = ( Xt-1 + Xt + Xt+1 + Xt+2 ) / 4  */
    
32. proc expand data=have method=none out=want4(drop=time);
    
33. by group notsorted;
    
34. convert value=value4 / transform=(cmovave 4 trimleft 1 trimright 2);
    
35. run;
    
36. /*3.2.2 one more lagged value than lead value is included in the time window
    
37. Yt = ( Xt-2 + Xt-1 + Xt + Xt+1 ) / 4  */
    
38. proc expand data=have method=none out=want5(drop=time);
    
39. by group notsorted;
    
40. convert value=value5 / transform=(reverse cmovave 4 reverse trimleft 2 trimright 1);
    
41. run;
    
42. 
    
43. /*3.2.3 weighted moving time window operator; the lead value and lagged value are all weighted 0.5
    
44. Yt = ( 0.5 * Xt-2 + Xt-1 + Xt + Xt+1 + 0.5 * Xt+2 ) / 4*/
    
45. proc expand data=have method=none out=want6(drop=time);
    
46. by group notsorted;
    
47. convert value=value61 / transform=(cmovave (0.5 1 1 1 0.5) trim 2 *5/4);
    
48. run;

复制代码

方法2：

复制代码

1. %macro move_avg(var,num);
   
2. %let lagnum=%eval(&num-1);
   
3. (&var
   
4. %do i=1 %to &lagnum;
   
5. +lag&i(&var)
   
6. %end;
   
7. )/&num
   
8. %mend move_avg;
   
9. 
   
10. /*前面5天的ltt均值作为当天的mean，只选取2008年来的数据*/
    
11. data r4(where=(m>=(&t1.+1)));
    
12. set r3;
    
13. aa=%move_avg(ltt,&t1.);/*这个参数可以改动*/
    
14. mean=lag(aa);
    
15. run;

复制代码

相关帖子：

计算5期移动平均的程序？
本文来自: 人大经济论坛 SAS专版 版，详细出处参考：http://www.pinggu.org/bbs/viewthread.php?tid=930483&page=1&from^^uid=1832606

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏1 回帖

关键词：移动平均 transform calculate operator weighted SAS

mark mark~ 1# 论坛数据分析

mark~ 1# 论坛数据分析