自我愚见，望有错指出改之。

/*SQL 学习*/

/*数据定义语言(DDL)：create、drop、alter*/

/*数据操作语言(DML)：insert、update、delete*/

/*数据查询语言(DQL)：select*/

/*数据控制语言(DCL)：grant、remove、commit、rollback*/

/*SELECT语句*/

proc sql；

select ...

from..

where..

group by..

having..根据group

by得到子数据集

order by..;

/*EX1*/

data a;

set sashelp.class;

where a<70;

run;

proc sql;

select sex='F'

from sashelp.classfit;

quit;

proc sql;

select sex='F'

from sashelp.classfit;

quit; /*s输出的结果是111000.... 表示如果是符合sex=‘F’的就是1*/

/*利用SQL创建新表*/

proc sql;

create table

new(表的名字) as

select * (代表选择全部)

from ..;

quit;

/*利用列创建新变量*/

proc sql;

create table

new as

select*,

weight/height as ratio /*这个蛮有意思

ratio是新的列名*/

format=5.2 label='体重：身高比'

from

sashelp.classfit;

quit;

/*利用SQL进行数据统计 类似Proc

summary */

proc sql;

create table

new1 as

select min(age) as youngest,

max(age) as oldest,

mean(height) as h format=5.1

from sashelp.classfit;

quit;

/*等同于下面的语句*/

proc summary data=sashelp.classfit;

var age height;

output out

=new2(drop=_type_ _freq_)/*删去两个列的统计量*/

min(age)=yougest

max(age)=oldgest

mean(height)=h; /*记住分号别忘了*/

run;

/*利用SQL进行分组统计 加入了group

by*/

proc sql;

create table

new3 as

select sex,

min(age) as youngest,

max(age) as oldest,

mean(height) as h format=5.1

from sashelp.classfit

group by

sex;

quit;

/*等同于下面的语句*/

proc summary data=sashelp.classfit;

class sex;

var age height;

output out

=new4(drop=_type_ _freq_)

min(age)=yougest

max(age)=oldgest

mean(height)=h;

run;

/*根据条件语句创建数据集

CASE WHEN ..THEN*/

proc sql;

create table

new5 as

select name,age,sex,

case when

age=11 then 'zoo'

when sex='F' then 'man'

else '[none]'

end

as nenene /*是对新的 zoo man 这列命名*/

from new;

quit;

/*利用where创建子数据集 简单*/

proc sql;

create table

new6 as

select *

from new

where sex='F';

quit;

/*利用having从句代替where从句

*/

proc sql;

create table

new7 as

select*

from new

group by

sex,age

having sex='F';

quit;

/**必记：having只能在group

by 之后 where必须在其之前**/

/*利用distinct剔除重复记录 **这个有陷阱见笔记*/

proc sql;

create table

new8 as

select distinct*

from new;

quit; /*这个例子看不出效果 是以行为标准*/

/*****************************************************/

/*SQL提升*/

/*一.将原始变量和二次分析生成的变量进行合并*/

proc sql;

create table

new9 as

select name,sex,

count(*) as

many label='统计'

from

new

group by age

order by name ;

quit;

proc sql;

create table

new10 as

select name,sex,

count(distinct

sex) as many label='统计'

from

new

group by age

order by name ;

quit; /** Really 看不懂这的结果

为什么输出来那样？？？怎么做到的统计

哦哦 看明白了 是利用年龄的统计 后面distinct age 剔除重复的年龄再统计*/

/*二.SQL语言的嵌套*/

proc sql;

create table

new11 as

select *

from(select

age, /*标点*/

count(*)as

many

from new

group by age

)

having many=max(many);/*重点*/

quit;

/*下面是不嵌套的*/

proc sql;

create table

new12 as

select age,

count(*)as

many

from new

group by age;

proc sql;

create table

new13 as

select *

from new12

having many=max(many);

quit;

/*三、连接 难难难~~~~*/

/*简单合并用merge在data步*/

/*交叉合并

cross join 、 full join(key)、nature

full join(key)

inner join、left

join、right join笔记本上*/

data one;

do a=1,2;

output;

end;

run;

data two;

do b=10,9,8;

output;

end;

run;

data combined;

merge one two;

run;

proc sql;

create table

combined as

select *

from one cross join two;

quit; /*两种合并处理效果不一样！！！*/

/*四、where exists、where in(where any)查询数据库中已有的观测*/

data H;

input name $age;

cards;

John 12

;

proc sql;

create table

HH as

select *

from H

where exists

(select*

from new

where h.name=new.name

);

quit; /*有趣哈哈哈哈*/

proc sql;

create table

HHH as

select *

from h

where name in

(select name

from new

);

quit;

proc sql;

create table

HHHH as

select *

from h

where name=any /*补充ANY用法 >any(20,30)相当于>20

(select name

from new

);

quit;

/*五、intersect(交集)，union(并集)，expect(补集)，

余集(得另外编程没有特定的代码)、union all(全集)、的区别*/

/*补充ALL用法 >all(20,30)相当于大于30，

/*any就是大于小的小于大的，all就是小于小的大于大的*/

/*小结：对小型表连接 数据步更有效，其余肯定SQL更有效啦*/

/*六、去掉标题(title；)，显示程序效率 (stimer)*/

TITLE;

proc sql feedback stimer;

select *

from new

order by

age;

reset nostimer outobs=3;

select *

from new

order by

age;

quit; /*厉害这样显示*/

/*SQL在几个主要程序测试与效率

exec/noexec：控制SQL语句是否在执行

nostimer/stimer：在sas日志中报告每个SQL语句的效率统计数据

noerrorstop/errorstop：批处理时控制当有错误发生时，是否要检查语法

*/

/*七、利用宏变量*/

data w;

input ID $ a1-a4;

cards;

A 11 12 13 14

B 21 22 23 24

;

/*计算各个变量之和*/

proc sql;

create table

sums as

select sum(a1) as sum1,

sum(a2) as sum2,

sum(a3) as sum3,

sum(a4) as sum4 /*记住这没有点号！！！*/

from W

;

quit;

/*等价*/

proc summary data=w;

var a1-a4;

output out=sumss(drop=_type_

_freq_)

sum=sum1-sum4;

run;

data sumsss;

set w end=last;

ARRAY _a{*}a1-a4; /*定义数组*/

ARRAY _sum{*}sum1-sum4;

keep sum1-sum4;

do i=1 to 4;

_sum{i]+_a{i};

end;

if last then output;

run;

/*变量更多时 就用宏*/

%macro selectsums(maxindex=);

%do n=1 %to &maxindex;

sum(a&n)

as sum&n

%if &n NE &maxindex %then ,

;

%END;

%mend selectsums;

PROC SQL;

create table

sumssss as

select %selectsums(maxindex=4)

from w

;

quit; /*完美！*/

/*八、利用SQL生成报表*/

/*九、混合复杂操作*/

/*行列转置 常规的转置有proc transpose ..*/

proc sql;

create table

ww as

select id,'a1' as item,

a1 as vlaue

from w

union all

select id,'a2' as item,

a2

from w

union all

select id,'a3' as item,

a3

from w

union all

select id,'a4' as item,

a4

from w;

quit;

proc sql;

create table

verticalsums as

select *

from ww;

group by

item;

quit;

by tt.