IML編程的基本函數是怎樣的

這篇文章將為大家詳細講解有關IML編程的基本函數是怎樣的，文章內容質量較高，因此小編分享給大家做個參考，希望大家閱讀完這篇文章后對相關知識有一定的了解。

SAS/IML 是利用矩陣語言的模塊，矩陣用于統計計算非常方便，典型代表是多元線性模型的求解。矩陣中行代表著觀測，列可以表示變量。字符型變量和數值型變量通吃。下面是IML基本函數,當然只是一小部分。。

矩陣運算函數

proc iml; /*聲明iml模塊*/
  x=1;
  y={1 2 3};
  z={1 2 3,
     3 5 6};
  o={'abc' 'opq',
     'lmn' 'rst'}; 
  n_z=nrow(z);/*矩陣維度，行列維度nrow(),ncol()*/
  p_o=ncol(o);
  type_o=type(o);/*type()矩陣類型，系字符型C或數值型N或者都包括*/
  nlen=nleng(o);/*nleng矩陣長度，幾個字符串*/
  len=length(o); /*length()矩陣中每個字符串的字符,當然字符串長度可以更改*/
  print x,y,z,o n_z; /*顯示矩陣值*/
quit;

利用函數創建矩陣

proc iml;
   c=j(10,1,3.14); /*J（）函數產出常量矩陣，10行1列3.14*/
   r=j(1,5);    /*一行五列的矩陣，默認值1*/
   m=j(10,5,0); /*10行5列的0矩陣*/
   miss=j(3,2,.); /*3行2列缺失值矩陣*/
   g=repeat({0 1},3,2);/*REPEAT()函數可以產生更一般的矩陣*/
   j=1:5;/*“ ：”標記從1到5，默認增量為1*/
   k=do(1,10,2);/*DO()函數可以進行算術計算，產生固定間距的矩陣*/
   seed=j(10,1,1);
   need=uniform(seed);/*生成偽隨機的矩陣，seed為1*/
   /*如果需要生成大量的偽隨機數，考慮RANDGEN程序*/
   call randseed(123);/*種子*/
   x=j(10,1); /*x系一個列向量*/
   e=x;       /*x賦值給e*/
   call randgen(x,"Uniform"); /*生成隨機向量*/
   call randgen(e,"Normal");
   y=3*x+2+e;                 /*回歸模型*/
print c,r,m,miss,need,y;

quit;
proc iml;
   s={1 2 3,4 5 6,7 8 9,10 11 12};
   transpose=t(s); /*矩陣轉置,T()函數完成*/
   s2=shape(s,6);/*SHAPE()函數用于矩陣重組，s2是6行矩陣*/
   s3=s[3,2];  /*從矩陣中提取某個值，找到其行列數即可完成*/
   s4=s[1,2];
   s5=s[s4,s4]; /*在提取數基礎上可以重新構造矩陣*/
   s6=vecdiag(s);/*vecdiag,可以求矩陣的對角陣*/
print transpose,s2,s3,s4,s5,s6 ;quit;

iml進行數據管理的函數

APPEND               adds observations to the end of a SAS data set
CLOSE                closes a SAS data set
CREATE               creates and opens a new SAS data set for input and output
DELETE               marks observations for deletion in a SAS data set
EDIT                 opens an existing SAS data set for input and output
FIND                 finds observations
INDEX                indexes variables in a SAS data set
LIST                 lists observations
PURGE                purges all deleted observations from a SAS data set
READ                 reads observations into IML variables
REPLACE              writes observations back into a SAS data set
RESET DEFLIB         names default libname
SAVE                 saves changes and reopens a SAS data set
SETIN                selects an open SAS data set for input
SETOUT               selects an open SAS data set for output
SHOW CONTENTS        shows contents of the current input SAS data set
SHOW DATASETS        shows SAS data sets currently open
SORT                 sorts a SAS data set
SUMMARY              produces summary statistics for numeric variables
USE                  opens an existing SAS data set for input

簡單例子：

data class;/*work庫下建立數據class*/
   set sashelp.class;
run;

/*下面是函數的簡單運用*/

   use class;
   show datasets contents; 
   reset deflib=work;

   m={3 6 9 13};
   n={name sex age weight};
   list all point m var n where(sex='F');

   use class;
   read all var {name height};
   read all var _num_ into x;
   read all var _num_ into femal where(sex="F");
   print x,female;

   edit class;
   find all where(name={'HENRY'}) into d;
   delete point=d;
   delete all where(age>12);
   
   create class2 from class;
   setin class2 point 0;
   sum=0;
   do data;
      read next var(weight);
   sum=sum+weight;
   end;
   print sum;

   create ratio2 var(name htwt);
   append;
   show contents;
   close ratio2;
   
   summary var {'height' 'weight'} class {sex} stat {mean std} opt{save};
   
   sort class out=sorted by name;
   
   index sex; 
   purge;

簡單的邏輯編程

proc iml;
   a={17 22,13 10};
   if max(a)<20 then p=1;
   else p=0;
print a,p;
quit;
proc iml;
   y=0;
   do i=1 to 10 by 2;
      y=y+i;
   end;
print y;
quit;
proc iml;
   count=1;
   do while (count<5);
      count=count+1;
   end;
print count;
quit;
proc iml;
   count=1;
   do until(count>5);
      count=count+1;
 end;
print count;
quit;
/* module without arguments, all symbols are global. */
proc iml;
   a = 10; /* a is global */
   b = 20; /* b is global */
   c = 30; /* c is global */
   start Mod1; /* begin module */
      p = a+b; /* p is global */
      c = 40; /* c already global */
   finish; /* end module */
   run Mod1; /* run the module */
print a b c p;

描述性統計量計算

Sums:
VEC_SUM: The sum of the elements of the first row of the matrix X
COL_TOT: The column totals (sum of the elements in each column of the matrix X
MAT_SUM: The sum of the elements of the matrix X
SUBMAT: A submatrix including the first two columns and first two rows of the matrix X
SM_SUM: The sum of the elements of a submatrix including the first two rows and the first two columns of the matrix X

PROC IML;
   x={1 2 3,4 5 6,7 8 9};
   vec_sum=x[1,+];
   col_tot=x[+,];
   mat_sum=SUM(x);
   submat=x[1:2,1:2];
   sm_sum=SUM(submat);
   PRINT vec_sum col_tot mat_sum sm_sum;
QUIT;

Means:
VEC_MEAN: The mean of the elements of the first row of the matrix X
MAT_MEAN: The mean of the elements of the matrix X
SM_MEAN: The mean of the elements of a submatrix including the first two rows and the first two columns of the matrix X

PROC IML;
   x={1 2 3,4 5 6,7 8 9};
   vec_mean=x[1,:];
   mat_mean=x[:];
   submat=x[1:2,1:2];
   sm_mean=submat[:];
   PRINT vec_mean mat_mean sm_mean;
QUIT;

Variances:
VEC_VAR: The variance of the elements of the first row of the matrix X
MAT_VAR: The variance of the elements of the matrix X
SM_VAR: The variance of the elements of a submatrix including the first two rows and the first two columns of the matrix X

PROC IML;
   x={1 2 3,4 5 6,7 8 9};
   ssq=SSQ(x[1,]);
   sum=x[1,+];
   vec_var=(ssq-sum*sum/NCOL(x))/(NCOL(x)-1);
   jn=J(NROW(x));
   mat_mean=x[:]; 
   mat_var=SSQ(x-mat_mean*jn)/(NROW(x)*NCOL(x)-1);
   submat=x[1:2,1:2];
   ssq_sub=SSQ(submat);
   sum_sub=SUM(submat);
   sm_var=(ssq_sub - sum_sub*sum_sub/(NCOL(submat)*NROW(submat))) /
           (NCOL(submat)*NROW(submat)-1);
   PRINT vec_var mat_var sm_var;
QUIT;

Other basic statistics:

PROC IML;
   scores={90,62,66,68,70,72,73,74,78,78,78,79,80,81,82,
           82,82,84,84,85,85,85,85,87,88,89,89,89,89,61};
   PRINT scores;
   count=NROW(scores);
   sum=SUM(scores);
   mean=sum/count; /* =scores[:] */
   median=MEDIAN(scores);
/** Frequency Distribution **/
START freq(x);
   unique=UNION(x)`; /** vector of unique scores **/
   observed=J(NROW(unique),1);
   DO i=1 TO NROW(unique);
      observed[i]= NCOL(LOC(x=unique[i]));
   END;
   freq=unique||observed; /* vector of unique scores and their frequencies */
   f_Label={" SCORE" "FREQUENCY"}; /* labels for the columns of Freq */
   PRINT "Frequency Distribution" freq [COLNAME=f_label];
FINISH;
CALL FREQ(scores);
/** End Frequency Distribution**/

/** Start Mode **/
START mode(x);
   unique=UNION(x)`;
   observed=J(NROW(unique),1);
   DO i=1 TO NROW(unique);
      observed[i]= NCOL(LOC(x=unique[i]));
   END;
   max=MAX(observed);
   mode=unique[LOC(observed=max),1];
   RETURN(mode);
FINISH;
mode=MODE(scores);
/** End Mode **/

ssq=SSQ(scores);
variance=(ssq-sum*sum/count)/(count-1);
stdev=SQRT(variance);
quartiles=QUARTILE(scores);

/** Start Percentile **/
START percentile(x,pct);
   CALL SORT(x,1);
   count=NROW(x);
   pct=pct/100*(count+1);
   fpct=FLOOR(pct);
   cpct=CEIL(pct);
   IF fpct=cpct THEN percentile=x[pct];
   ELSE percentile=x[fpct]+(pct-fpct)* (x[cpct]-x[fpct]);
/* SAS PCTLDF 4 */
RETURN(percentile);
FINISH;
percentile=PERCENTILE(scores,75);
/** End Percentile **/


CALL SORT(scores,1); /* sorts the scores in ascending order */
quantile=scores[CEIL(count*.45)];
PRINT count sum mean median mode,
      variance stdev quartiles, "75th Percentile:"
      percentile, "45th Quantile:" quantile;
QUIT;

IML簡單運用，相關系數的計算

ods trace output;
PROC IML;
/* Module to compute correlations */
start corr;
   n = nrow(x); /* number of observations */
   sum = x[+,] ; /* compute column sums */
   xpx = t(x)*x-t(sum)*sum/n; /* compute sscp matrix */
   s = diag(1/sqrt(vecdiag(xpx))); /* scaling matrix */
   corr = s*xpx*s; /* correlation matrix */
   print "Correlation Matrix",,corr[rowname=nm colname=nm] ;
finish corr;
/* Module to standardize data */
start std;
   mean = x[+,] /n; /* means for columns */
   x = x-repeat(mean,n,1); /* center x to mean zero */
   ss = x[##,] ; /* sum of squares for columns */
   std = sqrt(ss/(n-1)); /* standard deviation estimate*/
   x = x*diag(1/std); /* scaling to std dev 1 */
   print ,"Standardized Data",,X[colname=nm] ;
finish std;
/* Sample run */
x = { 1 2 3,3 2 1,4 2 1,0 4 1,24 1 0,1 3 8};
nm={age weight height};
run corr;
run std;

關于IML編程的基本函數是怎樣的就分享到這里了，希望以上內容可以對大家有一定的幫助，可以學到更多知識。如果覺得文章不錯，可以把它分享出去讓更多的人看到。