[原创博文] 如何提高sas的data step的读取速度 [推广有奖]

21楼

shewenhao 发表于 2011-1-21 02:05:21

18# dchrenstu
感觉这个相对强度还是蛮重要的，因为你会发现日内扰动往往会随大盘而变化，但是如果观察k线，会发现一些股票如果相对大盘比较强势，往往会后市走强的概率蛮高。所以需要统计上进一步的研究。你看呢？

22楼

myzhang1982

发表于 2011-1-24 09:25:40

16# shewenhao
我的不是多个dataset的问题，是只有一个很大的dataset，读入要花好长时间

23楼

情迷仲夏夜 发表于 2011-1-24 09:30:37

上传一段程序，也许会对你有所帮助：

%global libname dsn1 dsn2 dsn3 srcfile;

data fin;
input libname $ dsn1 $ dsn2 $ dsn3 $ srcfile $;
cards;
fy97v6 cen_fy97 cen97b cen97c cen97
fy97v6 cxm97 cxm97b cxm97c cxm97
fy97v6 ptf_fy97 ptf97b ptf97c ptf97
fy97v6 xm97 xm97b xm97c xm97
fy98v6 cen_98 cen98b cen98c cen98
fy98v6 excen_98 excen98b excen98c excen98
fy98v6 ptf_98 ptf98b ptf98c ptf98
fy98v6 xm98 xm98b xm98c xm98
fy98v6 obs_98 obs98b obs98c obs98
fy98v6 ext_98 ext98b ext98c ext98
fy98v6 cenobs98 cenobs98b cenobs98c cenobs98
run;

%macro dxfinptobs(libname, dsn1, dsn2, dsn3, srcfile);
data &dsn2(keep=scrssn dxp v1-v10 srcfile dm);
   set &libname..&dsn1(obs=max);
   array aa dxp v1-v10;
   do over aa;
      if aa in ('250', '3572', '3620', '36641') then dm=1;
   end;
   srcfile="&srcfile";

data &dsn3;
   set &dsn2;
   if dm=1;
run;
%mend dxfinptobs;

%macro dsn;
%do i=1 %to 11;
   data _null_;
      set fin;
      if _n_=&i then call symput('libname', libname);
         put libname=&libname;
      if _n_=&i then call symput('dsn1', dsn1); put dsn1=&dsn1;
      if _n_=&i then call symput('dsn2', dsn2); put dsn2=&dsn2;
      if _n_=&i then call symput('dsn3', dsn3); put dsn3=&dsn3;
      if _n_=&i then call symput('srcfile', srcfile);
   run;
   %dxfinptobs(&libname, &dsn1, &dsn2, &dsn3, &srcfile);
%end;
%mend dsn;

%dsn;

24楼

情迷仲夏夜 发表于 2011-1-24 09:39:36

也许还有另一种办法，这个办法将所有文件是按照年限，先读成几个文件，如果需要，再继续合并成一个大文件：
%let vars=id date;

data sf97 sf98 sf99 sf00 sf01;
id=.;
run;

%macro readsf(yr, infile);
%if &yr=97 %then
   %do;
      data a;
         set fy97sf.&infile(keep=&vars);

      proc sort; by id;

      data sf97;
         set sf97 a; by id;
         if id=. then delete;
      run;
   %end;
%else %if &yr=98 %then
   %do;
      data a;
         set fy98sf.&infile(keep=&vars);

      proc sort; by id;

      data sf98;
         set sf98 a; by id;
         if id=. then delete;
      run;
   %end;
%else %if &yr=99 %then
   %do;
      data a;
         set fy99sf.&infile(keep=&vars);

      proc sort; by id;

      data sf99;
         set sf99 a; by id;
         if id=. then delete;
      run;
   %end;
%else %if &yr=00 %then
   %do;
      data a;
         set fy00sf.&infile(keep=&vars);

      proc sort; by id;

      data sf00;
         set sf00 a; by id;
         if id=. then delete;
      run;
   %end;
%else %if &yr=01 %then
   %do;
      data a;
         set fy01sf.&infile(keep=&vars);

      proc sort; by id;

      data sf01;
         set sf01 a; by id;
         if id=. then delete;
      run;
   %end;
%mend readsf;

%readsf(97, sf97v0102) %readsf(97, sf97v0304) %readsf(97, sf97v0506)
%readsf(97, sf97v0708) %readsf(97, sf97v0910) %readsf(97, sf97v1112)
%readsf(97, sf97v1314) %readsf(97, sf97v1516) %readsf(97, sf97v1718)
%readsf(97, sf97v1920) %readsf(97, sf97v2122)

%readsf(98, sf98_1) %readsf(98, sf98_2) %readsf(98, sf98_3)
%readsf(98, sf98_4) %readsf(98, sf98_5) %readsf(98, sf98_6)
%readsf(98, sf98_7) %readsf(98, sf98_8) %readsf(98, sf98_9)
%readsf(98, sf98_10) %readsf(98, sf98_11)

%readsf(99, sf990102) %readsf(99, sf990304) %readsf(99, sf990506)
%readsf(99, sf990708) %readsf(99, sf990910) %readsf(99, sf991112)
%readsf(99, sf991314) %readsf(99, sf991516) %readsf(99, sf991718)
%readsf(99, sf991920) %readsf(99, sf992122)

%readsf(00, sf00v0107) %readsf(99, sf00v0814)
%readsf(00, sf00v1518) %readsf(00, sf00v1922)

%readsf(01, sf010105) %readsf(01, sf010608) %readsf(01, sf010911)
%readsf(01, sf011214) %readsf(01, sf011517) %readsf(01, sf011820)
%readsf(01, sf012122)