兩種SAS代碼實現(xiàn)變量的缺失值頻數(shù)及占比

sas對缺失值的統(tǒng)計，可得出缺失值的頻數(shù)及占比。以下為詳細程序代碼：
/*首先是創(chuàng)建示例數(shù)據(jù)集*/
data tmp;
infile datalines delimiter=",";
length var1 $8.;
length var2 8.;
length var3 $8.;
length var4 8.;
input var1 $ var2 var3 $ var4
@@;
datalines;
A,1,B,2.,3,C,.C,.,.,3
;
run;

如圖得到下列數(shù)據(jù)集

然后統(tǒng)計數(shù)據(jù)集中缺失值和占比，先為字符型和數(shù)值型先分別設(shè)定一個format，然后直接對變量進行頻數(shù)統(tǒng)計，再做一下簡單處理，就可得到理想結(jié)果。

代碼如下

proc format;

value num_f . = "0"

low-high = "1" ;

value $char_f " " = "0"

other = "1" ;

run;

/*頻數(shù)統(tǒng)計*/

ods output onewayfreqs=tables;

proc freq data= tmp ;

tables _all_ / missing;

format _numeric_ num_f. _character_ $char_f.;

run;

ods output close;

數(shù)據(jù)集如下：

/*保留缺失變量、頻數(shù)和占比*/

data miss;

length variable $50;

set tables;

variable = scan(Table,2,"“"); /*獲取變量名*/

/*由于變量都是F_開頭，因此可以用F_:來包含所有變量*/

value = max(of F_:);

if value = 0; /*缺失標志*/

keep variable frequency percent;

label variable = "缺失變量名" frequency = "缺失頻數(shù)" percent = %nrstr("%缺失占比");

run;

以上為第一種方法；

下面用數(shù)組的方法進行實現(xiàn)。

/*找出缺失變量*/

data tmp11;

set tmp;

array arr1{*} _NUMERIC_ ;

array arr2{*} _CHARACTER_ ;

length variable $50;

do i = 1 to dim(arr1);

if missing(arr1(i)) then do;

variable =vname(arr1(i)); /*數(shù)值型缺失*/

output;

end;

end;

do j = 1to dim(arr2);

if missing(arr2(j)) then do;

variable = vname(arr2(j)); /*字符型缺失*/

output;

end;

end;

keep variable;

run;

/*統(tǒng)計缺失頻數(shù)和占比*/

proc sql noprint;

select count(*) into : N from tmp;

create table miss as

select variable label = "缺失變量名",count(*) as frequency label = "缺失頻數(shù)",

input(compress(put(calculated frequency / &N.,percent10.2),"%"),best32.) as percent label = %nrstr("%缺失占比")

from tmp11

group by variable;

quit;