Oracle中的[行转列]与[列转行]

一、原始数据

二、行转列的多种实现方式

[1.CASE WHEN](#1.CASE WHEN)

2.DECODE

[2.UNION ALL合并结果集](#2.UNION ALL合并结果集)

[四、行转列练习：CASE WHEN/DECODE/PIVOT/lag/LEAD](#四、行转列练习：CASE WHEN/DECODE/PIVOT/lag/LEAD)

[1.CASE WHEN](#1.CASE WHEN)

2.DECODE

3.PIVOT

4.LEAD

一、原始数据

sql 复制代码

CREATE TABLE T_SCORE (SNO NUMBER,CLA VARCHAR2(20),SCORE NUMBER);

INSERT INTO T_SCORE VALUES (101,'语文',88);
INSERT INTO T_SCORE VALUES (101,'数学',89);
INSERT INTO T_SCORE VALUES (101,'英语',90);
INSERT INTO T_SCORE VALUES (102,'语文',91);
INSERT INTO T_SCORE VALUES (102,'数学',77);
INSERT INTO T_SCORE VALUES (102,'英语',56);
INSERT INTO T_SCORE VALUES (103,'语文',77);
INSERT INTO T_SCORE VALUES (103,'数学',88);
INSERT INTO T_SCORE VALUES (103,'英语',99);
INSERT INTO T_SCORE VALUES (104,'语文',77);
INSERT INTO T_SCORE VALUES (104,'数学',66);
INSERT INTO T_SCORE VALUES (104,'英语',55);
INSERT INTO T_SCORE VALUES (105,'语文',44);
INSERT INTO T_SCORE VALUES (105,'数学',67);
INSERT INTO T_SCORE VALUES (105,'英语',78);
INSERT INTO T_SCORE VALUES (106,'语文',89);
INSERT INTO T_SCORE VALUES (106,'数学',98);
INSERT INTO T_SCORE VALUES (106,'英语',78);
INSERT INTO T_SCORE VALUES (107,'语文',67);
INSERT INTO T_SCORE VALUES (107,'数学',56);
INSERT INTO T_SCORE VALUES (107,'英语',54);
INSERT INTO T_SCORE VALUES (108,'语文',76);
INSERT INTO T_SCORE VALUES (108,'数学',78);
INSERT INTO T_SCORE VALUES (108,'英语',12);
COMMIT;

SELECT * FROM T_SCORE;

二、行转列的多种实现方式

原格式：

行转列后的格式：

1.CASE WHEN

sql 复制代码

select SNO
     , CASE WHEN CLA = '语文' THEN SCORE end as 语文
     , CASE WHEN CLA = '数学' THEN SCORE end as 数学
     , CASE WHEN CLA = '英语' THEN SCORE end as 英语
from T_SCORE;

下面两种聚合函数都可以：

sql 复制代码

select SNO
     , max(CASE WHEN CLA = '语文' THEN SCORE end) as 语文
     , max(CASE WHEN CLA = '数学' THEN SCORE end) as 数学
     , max(CASE WHEN CLA = '英语' THEN SCORE end) as 英语
from T_SCORE
group by SNO
order by SNO;

select SNO
     , sum(CASE WHEN CLA = '语文' THEN SCORE end) as 语文
     , sum(CASE WHEN CLA = '数学' THEN SCORE end) as 数学
     , sum(CASE WHEN CLA = '英语' THEN SCORE end) as 英语
from T_SCORE
group by SNO
order by SNO;

二者区别：

聚合函数	逻辑	处理重复记录	适用场景
`MAX`	返回分组内的最大值。若每组只有一个值，则直接返回该值。	保留最大值（如补考成绩）。	行转列（提取唯一值）。
`SUM`	返回分组内所有值的总和。若每组只有一个值，则返回该值本身。	累加所有值（可能导致成绩异常）。	统计总分或合计。

2.DECODE

sql 复制代码

select SNO
     , decode(CLA, '语文', SCORE) as 语文
     , decode(CLA, '数学', SCORE) as 数学
     , decode(CLA, '英语', SCORE) as 英语
from T_SCORE;

sql 复制代码

select SNO
     , max(decode(CLA, '语文', SCORE)) as 语文
     , max(decode(CLA, '数学', SCORE)) as 数学
     , max(decode(CLA, '英语', SCORE)) as 英语
from T_SCORE
group by SNO
order by SNO;

select SNO
     , sum(decode(CLA, '语文', SCORE)) as 语文
     , sum(decode(CLA, '数学', SCORE)) as 数学
     , sum(decode(CLA, '英语', SCORE)) as 英语
from T_SCORE
group by SNO
order by SNO;

3.PIVOT(Oracle独有)

语法：

sql 复制代码

PIVOT (SUM(聚合值) FOR 待转换的列名 IN (待转换的列名里面的值 转换后列的别名))

sql 复制代码

select *
from T_SCORE
    PIVOT (sum(SCORE) for CLA in ( '语文' Chinese,'数学' Math,'英语' English))
order by SNO;

4.使用LEAD开窗函数

因为要对中文进行排序，所以先使用ASCII码

sql 复制代码

SELECT CLA, ASCII(CLA)
FROM T_SCORE
GROUP BY CLA
ORDER BY CLA;

sql 复制代码

SELECT *
FROM (SELECT sno
           , LEAD(score, 0) OVER (PARTITION BY sno ORDER BY CLA ) 数学
           , LEAD(score, 1) OVER (PARTITION BY sno ORDER BY CLA ) 英语
           , LEAD(score, 2) OVER (PARTITION BY sno ORDER BY CLA ) 语文
      FROM T_SCORE)
WHERE 语文 IS NOT NULL;

三、列转行的多种实现方式

sql 复制代码

DROP TABLE b_score;
CREATE TABLE b_score AS
SELECT *
FROM (SELECT *
      FROM t_score
          PIVOT (
          SUM(score) -- 聚合函数（使用 SUM 或 MAX 均可）
          FOR cla IN (
              '语文' AS 语文, -- 指定课程名称及对应的列别名
              '数学' AS 数学,
              '英语' AS 英语
              )
          ))
ORDER BY sno; -- 按学生编号排序
COMMIT;
SELECT *
FROM b_score;

原格式：

行转列后的格式：

1.UNPIVOT(Oracle独有)

语法：

sql 复制代码

UNPIVOT (存储指标值的列名 FOR 合并后的列名 IN (待合并的列名))

sql 复制代码

SELECT *
FROM b_score
    UNPIVOT (score FOR cla IN (语文,数学,英语));

2.UNION ALL合并结果集

sql 复制代码

SELECT sno,'语文' cla,语文 score FROM b_score
UNION ALL
SELECT sno,'数学' cla,数学 score FROM b_score
UNION ALL
SELECT sno,'英语' cla,英语 score FROM b_score
order by SNO;

四、行转列练习：CASE WHEN/DECODE/PIVOT/lag/LEAD

输出每个部门的总人数，展示样式：

D10 D20 D30

3 5 6

1.CASE WHEN

sql 复制代码

select COUNT(case when DEPTNO = 10 then 1 end) as D10
     , COUNT(case when DEPTNO = 20 then 1 end) as D20
     , COUNT(case when DEPTNO = 30 then 1 end) as D30
from EMP;

2.DECODE

sql 复制代码

select COUNT(DECODE(DEPTNO, 10, 1)) as D10
     , COUNT(DECODE(DEPTNO, 20, 1)) as D20
     , COUNT(DECODE(DEPTNO, 30, 1)) as D30
from EMP;

3.PIVOT

sql 复制代码

SELECT DEPTNO
     , COUNT(EMPNO) CM
FROM EMP
GROUP BY DEPTNO;

sql 复制代码

select *
from (SELECT DEPTNO
           , COUNT(EMPNO) CM
      FROM EMP
      GROUP BY DEPTNO)
    pivot (sum(cm) for DEPTNO in (10 D10,20 D20,30 D30));

4.LEAD

sql 复制代码

select DEPTNO, count(empno) ct
from EMP
group by DEPTNO;

sql 复制代码

SELECT *
FROM (select lead(ct, 0) over ( order by DEPTNO) as D10
           , lead(ct, 1) over ( order by DEPTNO) as D20
           , lead(ct, 2) over ( order by DEPTNO) as D30
      from (select DEPTNO, count(empno) ct
            from EMP
            group by DEPTNO)) a
WHERE D30 IS NOT NULL;