能看看循环体或者是楼主的大致想法吗

我理解1维聚类的终止条件为所有CLUSTER的极差的最大值小于给定值，如果真是需要最小值的话，修改那个宏赋值语句max为min就可以了,不是统计出生，聚类也是半路出家，呵呵
三个参数依次为起始CLUSTER数，实验最大CLUSTER数，那个给定值。宏很粗糙，没做太多判别，稍微测试了下，可以用，希望对你有帮助。

祝新年快乐

Use %return; statement.

Here is the definition for %return

The %RETURN macro causes normal termination of the currently executing macro.



%macro test(dummy,test=out);
   %do i = 1 %to 5;
       %put >>>&i<<<;
       %if %upcase(&test)=OUT %then %return;
    %end;

%mend test;

%test(dummy,test=stay)
%test(dummy,test=out)

9楼的同志的程序是不是应该可以理解为退出%do，等价地退出宏。我仿照这个理解，编写如下程序：
%macro ml(n,m,zb);
data a;
do i=1 to &n;
x=uniform(0);
fi=arcos(-1);
y=sin(2*fi*x);
output;
keep y x fi i;
end;
%do j=2 %to &m;
proc fastclus data=a out=Clust&j maxclusters=&j noprint;
var x;
proc sort out=flei&j;by cluster;
proc means noprint data=flei&j;
var x; by cluster;
output out=menx&j(keep=rx) range=rx;
proc means data=menx&j noprint;
var rx;
output out=big&j max=zuizhi;
data pb&j;
set  big&j ;
%if zuizhi le &zb %then %return;
%end;
%%mend ml(n,m,zb);
%ml(200,15,0.4);run;
但提交出现问题，问题就在%if 这里，不知高手是否给予修正呢？
谢谢。
上面程序是确定一个最好的分类数，使得各个类内的变量的极差最大值小于事先给定的值，此时退出循环，得到最终的分类结果。

%return是直接终止宏了，而不是跳出循环。

%%mend ml(n,m,zb);
多了一个%。



data a;
        do i=1 to &n;
                x=uniform(0);
                fi=arcos(-1);
                y=sin(2*fi*x);
                output;
                keep y x fi i;
        end;


%do j=2 %to &m;

但愿可以

You are confused with macro programming and data step programming/SAS programming. A macro needs to compile before executed. A macro is nothing but a text generator.  Based on a data step variable to determine if exit-timing a macro is not working. Here is a workable idea.

Write two macro
1) A macro check the value to determine if it is needed to call another macro. In your case it will be in a data step.
2) Calculating macro in your case will be a proc fastclus, ect.

HTH

10# harlon1976


Based on what I understand your pseudo-macro codes I modify as below.
%let n=200;
data a;
do i=1 to &n;
x=uniform(0);
fi=arcos(-1);
y=sin(2*fi*x);
output;
keep y x fi i;
end;
run;
%macro fast(a, j);
proc fastclus data=a out=Clust&j maxclusters=&j noprint;
var x;
run;
proc sort out=flei&j;by cluster;
run;
proc means noprint data=flei&j;
var x; by cluster;
output out=menx&j(keep=rx) range=rx;
run;
proc means data=menx&j noprint;
var rx;
output out=big&j max=zuizhi;
run;
%mend;
%macro ml(m,zb);
%let trigger=0;
%do j=2 %to &m;
%if &trigger=0 %then %fast(a,&j);
data _null_;
set big&j;
if zuizhi le &zb then call symputx('trigger',1);
stop;
run;
%put >>>>>>>loop number j=&j;
%if &trigger=1 %then %return;
%if &j=&m %then %return;
%end;
%%mend ml;

%ml(20,0.1);

%GOTO 语句可以实现 参考如下：

%LET VAR=%STR(A B C);

%MACRO TEST;

%DO I=1 %TO 3;
        %LET T=%SCAN(&VAR.,&I);
        %PUT &T.;
        %IF "&T."="B" %THEN %DO;
                %LET C="HOLLE WORLD";
                %GOTO QUIT;
        %END;
%END;

%QUIT:

%MEND;       
       
%TEST;