行列转换实列

行列转换实例
                  表ttt有三个字段
                  seq  
--序列
                  jcxm 
--检查项目
                  zhi  
--

                  数据分别如下:
                  seq   jcxm       zhi
                  
-------      --------          --------
                  
11     1    0.50
                  
11     2    0.21
                  
11     3    0.25
                  
12     1    0.24
                  
12     2    0.30
                  
12     3    0.22                             

                  实现功能
                  创建视图时移动行值为列值


                  create view v_view1
                  
as
                  select seq,
                         sum(decode(jcxm,
1, zhi)) 检测项目1,
                         sum(decode(jcxm,
2, zhi)) 检测项目2, 
                         sum(decode(jcxm,
3, zhi)) 检测项目3 
                  from ttt
                  group by seq;

                  序号 检测项目1  检测项目2  检测项目3
                  
11     0.50    0.21     0.25
                  
12     0.24    0.30     0.22

 

                  技巧:
                  用THEN中的0和1来进行统计(SUM)

                  jcxm   zhi
                  
----   ----
                  a           
1
                  b           
1
                  a           
3
                  d           
2
                  e           
4
                  f           
5
                  a           
5
                  d           
3
                  d           
6
                  b           
5
                  c           
4
                  b           
3
                  求他的zhi既是1,也是3,也是5的jcxm
                  方法一
                  select jcxm
                  from ttt
                  group by jcxm
                  having sum(decode(zhi,
1,-1,3,-1,5,-1,0)) = -3
                  方法二
                  select jcxm from ttt 
                  group by jcxm having (sign(sum(decode(zhi,
1,-1,0)))+
                  sign(sum(decode(zhi,
3,-1,0)))+sign(sum(decode(zhi,5,-1,0)))<=-3);

                  
----------
                  a
                  b
                  说明:
                  sign()函数根据某个值是0、正数还是负数,分别返回0、
1-1
                  所以可以用sign和decode来完成比较字段大小来区某个字段
                  select decode(sign(字段1
-字段2),-1,字段3,字段4) from dual;

                  sign是一个对于写分析SQL有很强大的功能
                  下面我对sign进行一些总结:
                  但属性student取0和1以外的值,或者student取两个以上的标法值,问题就不会这么简单了
                  解决办法就是特征函数(abs(),sign())

                  常用的特征算法
                  [A=B]=
1-abs(sign(A-B))
                  [A!=B]=abs(sign(A-B)) 
                  [A
<B]=1-sign(1+sign(A-B)) 
                  不能用
-sign(A-B):因为如果不满足A<b则返回-1,而不是0,这样就不能用在字段选择上了
                  [A
<=B]=sign(1-sign(A-B))
                  [A
>B]=1-sign(1-sign(A-B))
                  [A
>=B]=sign(1+sign(A-B)))
                  [NOTα]=
1-d [α]
                  [αANDb ]=d [α]
*d [b ] (6)
                  [αOR b ]=sign(d [α]+d [b ])

                  例如:
                  A
<B                         Decode( Sign(A-B), -110 )      
                     
                  A
<=B                         Decode( Sign(A-B), 101 )      
                     
                  A
>B                         Decode( Sign(A-B), 110 )       
                    
                  A
>=B                         Decode( Sign(A-B), -101 )     
                      
                  A
=B                         Decode( A, B, 10 )         
                  A between B and C      Decode( Sign(A
-B), -10
                  Decode(Sign(A
-C), 101 ))         
                  A 
is null                       Decode(A,null,1,0)         
                  A 
is not null                 Decode(A,null,0,1)         A in 
                  (B1,B2,,Bn)  Decode(A,B1,
1,B2,1,,Bn,1,0)         
                  nor LogA                    Decode( LogA, 
010 )            
                    (
1-Sign(LogA)) 
                  LogA and LogB            LogA 
* LogB 
                  LogA or LogB              LogA 
+ LogB 
                  LogA xor LogB            Decode(Sign(LogA),Sign(LogB),
0,1)    
                  Mod(Sign(LogA),Sign(LogB),
2


                  
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

                  [NextPage]
                  另外一个关于成绩的分析例子

                  SELECT
                  SUM(CASE WHEN cj 
<60 THEN 1 ELSE 0 END) as "not passed",
                  SUM(CASE WHEN cj BETWEEN 
60 AND 79 THEN 1 ELSE 0 END) as 
                  
"passed",
                  SUM(CASE WHEN cj BETWEEN 
80 AND 89 THEN 1 ELSE 0 END) as 
                  
"good",
                  SUM(CASE WHEN cj 
>=90 THEN 1 ELSE 0 END) as "Excellent"
                  FROM cjtable;

                  decode用法2
                  表、视图结构转化
                  现有一个商品销售表sale,表结构为:
                  month    
char(6)      --月份
                  sell    number(
10,2)    --月销售金额

                  现有数据为:
                  
200001  1000
                  
200002  1100
                  
200003  1200
                  
200004  1300
                  
200005  1400
                  
200006  1500
                  
200007  1600
                  
200101  1100
                  
200202  1200
                  
200301  1300

                  想要转化为以下结构的数据:
                  year   
char(4)          --年份
                  
------------   ---------------------         
                  
-------------------
                  month1  number(
10,2)   --1月销售金额
                  month2  number(
10,2)   --2月销售金额
                  month3  number(
10,2)   --3月销售金额
                  month4  number(
10,2)   --4月销售金额
                  month5  number(
10,2)   --5月销售金额
                  month6  number(
10,2)   --6月销售金额
                  month7  number(
10,2)   --7月销售金额
                  month8  number(
10,2)   --8月销售金额
                  month9  number(
10,2)   --9月销售金额
                  month10  number(
10,2)     --10月销售金额
                  month11  number(
10,2)     --11月销售金额
                  month12  number(
10,2)     --12月销售金额

                  结构转化的SQL语句为:
                  create or replace view
                  v_sale(year,month1,month2,month3,month4,month5,month6,month7,month8,month9,month10,month11,month12)
                  
as
                      select 
                      substrb(month,
1,4),
                      sum(decode(substrb(month,
5,2),'01',sell,0)),
                      sum(decode(substrb(month,
5,2),'02',sell,0)),
                      sum(decode(substrb(month,
5,2),'03',sell,0)),
                      sum(decode(substrb(month,
5,2),'04',sell,0)),
                      sum(decode(substrb(month,
5,2),'05',sell,0)),
                      sum(decode(substrb(month,
5,2),'06',sell,0)),
                      sum(decode(substrb(month,
5,2),'07',sell,0)),
                      sum(decode(substrb(month,
5,2),'08',sell,0)),
                      sum(decode(substrb(month,
5,2),'09',sell,0)),
                      sum(decode(substrb(month,
5,2),'10',sell,0)),
                      sum(decode(substrb(month,
5,2),'11',sell,0)),
                      sum(decode(substrb(month,
5,2),'12',sell,0))
                      from sale
                      group by substrb(month,
1,4);

                  体会:要用decode 
/group by/ order by/sign/sum来实现不同报表的生成 
                  
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
                  CASE应用

                  
1        1        部门a        800        男
                  
2        2        部门b        900        女
                  
3        3        部门a        400        男
                  
4        4        部门d        1400        女
                  
5        5        部门e        1200        男
                  
6        6        部门f        500        男
                  
7        7        部门a        300        女
                  
8        8        部门d        1000        男
                  
9        9        部门d        1230        女
                  
10        10        部门b        2000        女
                  
11        11        部门c        2000        男
                  
12        12        部门b        1200        男

                    SELECT jcxm 
as 部门,COUNT(seq) as 人数,
                      SUM(CASE SEX WHEN 
1 THEN 1 ELSE 0 END) as 男,
                            SUM(CASE SEX WHEN 
2 THEN 1 ELSE 0 END) as 女,
                      SUM(CASE SIGN(zhi
-800) WHEN -1 THEN 1 ELSE 0 END) as 
                  小于800元,
                      SUM((CASE SIGN(zhi
-800)*SIGN(zhi-1000)                    
                  
/**//*用*来实现<和>功能*/
                           WHEN 
-1 THEN 1 ELSE 0 END)+(CASE zhi
                           WHEN 
800  THEN 1 ELSE 0 END)) as 从800至999,         
                  
/**//*注意别名不能以数字开头*/
                      SUM((CASE SIGN(zhi
-1000)*SIGN(zhi-1200)
                           WHEN 
-1 THEN 1 ELSE 0 END)+(CASE zhi
                           WHEN 
1000 THEN 1 ELSE 0 END)) as 从1000元至1199元,
                      SUM((CASE SIGN(zhi
-1200) WHEN 1 THEN 1 ELSE 0 END)
                      
+(CASE zhi WHEN 1200 THEN 1 ELSE 0 END)) as 大于1200元
                  FroM ttt 
                  GROUP BY jcxm

                  部门名 人数    男       女   小于800元 从800至999 从1000元至1199元   大于1200元
                  部门a        
3        2        1        2        1           0   
                                             
0
                  部门b        
3        1        2        0        1           0   
                                             
2
                  部门c        
1        1        0        0        0           0   
                                            
1
                  部门d        
3        1        2        0        0           1   
                                            
2
                  部门e        
1        1        0        0        0             0 
                                              
1
                  部门f        
1        1        0        1        0           0   
                                            
0
原文地址:https://www.cnblogs.com/shengel/p/765398.html