重新排序

Question

我有以下含关于每个组是如何解决问题的数据表：

PROB | GROUP | REPNO | STATUS 
-----+---------+-------+------- 
1 | Juniors | 0 | 
1 | Seniors | 1 | 
1 | Juniors | 2 | 
1 | Experts | 3 | SOLVED 
2 | Juniors | 0 | 
2 | Seniors | 1 | SOLVED 

柱PROB定义了基团中解决问题，列组定义组正在研究这一概率其中，柱REPNO定义解决重复的次数（重复尝试直到最终解决，0表示首先尝试，不重复），最后一列STATUS定义是否在该特定尝试中解决了任务。在这里，我可以通过所有群组（SELECT ... GROUP BY repno）在每次重复中有效地解决工作问题。

但我想说明probs是如何有效地解决每个特定的组（每组自己的重复顺序分布）。例如，PROB 1被小组试验了2次，并且一次被老年人组解决，最终在小组专家的第一次尝试中得到解决。

所以我需要重复的重新计算对于每个特定组：

PROB | GROUP | REPNO | REPNO_J | REPNO_S | REPNO_E | STATUS 
-----+---------+-------+---------+---------+---------+------- 
1 | Juniors | 0 |  0 |   |   | 
1 | Seniors | 1 |   |  0 |   |   
1 | Juniors | 2 |  1 |   |   |   
1 | Experts | 3 |   |   |  0 | SOLVED <-- experts solved in first try 
2 | Juniors | 0 |  0 |   |   | 
2 | Seniors | 1 |   |  0 |   | SOLVED <-- seniors solved in first try 

如何让这个重新计算？

Answer 1

使用RANK（）SQL FIDDLE

select 
    PROB, 
    GROUP1, 
    REPNO, 
    DECODE(GROUP1,'Juniors', rank() over (partition by prob, group1 order by repno) - 1) as REPNO_J, 
    DECODE(GROUP1,'Seniors', rank() over (partition by prob, group1 order by repno) - 1) as REPNO_S, 
    DECODE(GROUP1,'Experts', rank() over (partition by prob, group1 order by repno) - 1) as REPNO_E, 
    STATUS 
from mytable 
order by prob,repno 

Answer 2

可以使用CASE语句

SELECT "PROB", 
     "GROUP", 
     "REPNO", 
     (CASE WHEN "GROUP" = 'Juniors' THEN REPNO ELSE NULL END) as REPNO_J, 
     (CASE WHEN "GROUP" = 'Seniors' THEN REPNO ELSE NULL END) as REPNO_S, 
     (CASE WHEN "GROUP" = 'Experts' THEN REPNO ELSE NULL END) as REPNO_E 
FROM Table1 

Answer 3

使用ROW_NUMBER窗函数来获得每组人数：

select 
    prob, 
    "GROUP", 
    repno, 
    case when "GROUP" = 'Juniors' then try end as repno_j, 
    case when "GROUP" = 'Seniors' then try end as repno_s, 
    case when "GROUP" = 'Experts' then try end as repno_e, 
    status 
from 
(
    select 
    prob, "GROUP", repno, status, 
    row_number() over(partition by prob, "GROUP" order by repno) - 1 as try 
    from mytable 
) 
order by prob, repno; 

顺便说一句：这不是一个好主意，名称的列组。这是SQL中的保留字。因此，您必须使用引号并在查询中使用该名称时考虑大写/小写。