Hashing,the best tool for Searching

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

Searching is the most wildely used technology.It includes:merges,joins,formats,indexs and other special functions.
The skills about merge,sql and conditional logic are all base on comparsion.Because it need to compare key with one or more keys in another table,it is slow and it would spend a lot of memory.What is more,if you use merge,you have to sort the data first.If the data happen to be a large data,the Merge method would be the worst.
Hashing is better method base on Direct-Addressing and it is much more convenient.
The is a Demo using hash.You can try it in another way like sql,merge to compare their efficiency.

1. data a;
   
2. input id type $;
   
3. cards;
   
4. 1  a,b
   
5. 2  a
   
6. 3  a,b,c
   
7. ;
   
8. run;
   
9. 
   
10. data b;
    
11. input key $ data $;
    
12. cards;
    
13. a         xy
    
14. b         er
    
15. c         abc
    
16. run;

复制代码

Table A is the main dataset,The question is you need the replace the variable 'type' using table B.For example using 'xy' to replace 'a'.As a result,the first line would be:

1. 1 xy,er

复制代码

The solution I will post later.welcome disscusion.

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3 回帖

关键词：searching searchin search Ching Tool includes another compare formats special

很好

The 'merge' version:

1. data a;
   
2. input id type $;
   
3. cards;
   
4. 1 a,b
   
5. 2 a
   
6. 3 a,b,c
   
7. ;
   
8. data b;
   
9. input key $ data $;
   
10. cards;
    
11. a xy
    
12. b er
    
13. c abc
    
14. ;
    
15. run;
    
16. data out;
    
17. set a;
    
18. i=1;
    
19. do while(scan(type,i)^='');
    
20.    key=scan(type,i);
    
21.    i=i+1;
    
22.    output;
    
23. end;
    
24. run;
    
25. proc sort data=out;by  key;run;
    
26. data out;
    
27. merge out b;
    
28. by key;
    
29. keep id data i;
    
30. i=i-1;
    
31. run;
    
32. proc sort data=out;by id i;run;
    
33. data out;
    
34. length id 8 type $100;
    
35. retain type ;
    
36. set out;
    
37. by id i;
    
38. if first.id then type=compress(data);
    
39. else type=compress(type)||','||compress(data);
    
40. if last.id then output;
    
41. keep id type;
    
42. run;

复制代码

Professor, I'd pick the FORMAT method for your sample question because of the single key variable.

邓贵大 发表于 2013-5-24 11:09
Professor, I'd pick the FORMAT method for your sample question because of the single key variable.

Can you share the code of FORMAT method?

楼主的实力，可以开专栏了。

希望版主能邀请楼主开个SAS学习的小专栏，每隔几天给大家分享交流一些东西。
谢谢。

zhou.wen 发表于 2013-5-24 11:34
Can you share the code of FORMAT method?

No, you keep lecturing! I was just trying to bump up your post!
http://www.theprogrammerscabin.com/NS9PO13P.pdf
http://support.sas.com/resources ... ings09/071-2009.pdf

我也来试试

data _null_;
  set b;
  call symputx(key, data);
run;

data c;
  set a;
  length str $100;
  k = 1;
  str = "";
  do while (scan(type, k, ',') ne "");
    if k=1 then str = symget(scan(type, k, ','));
        else str =  strip(str)||","||symget(scan(type, k, ','));
        k = k + 1;
  end;
  drop k;
run;

ntsean 发表于 2013-5-24 13:52
我也来试试

data _null_;

Good try.
Using the macro variable.