Question

很抱歉Oracle-Noob的问题，但我遇到的问题是我正在尝试运行，而且似乎无法弄清楚如何编写它。

如果我将查询分成两部分，我会得到每个部分的正确信息。参见：

select   NVL (skill.category, 'Total:') "Skill Category",
         count (training.code) "# of Trainings"
from     skill join training on skill.code = training.code
group by rollup (skill.category);

返回：

Skill Category  # of Trainings
--------------- --------------
Database                     2
HR                           1
Leadership                   1
Printing                     1
Sales                        3
Web Design                   5
Total:                      13

同样，我得到以下第二个查询：

select NVL (skill.category, 'Total:') "Skill Category",
       count (project.code) "projects req training"
from   skill join project on skill.code = project.code
group by rollup (skill.category);

Skill Category  projects req training
--------------- ---------------------
Database                            2
Printing                            3
Web Design                          5
Total:                             10

然而，当我尝试组合这些查询时，我的结果搞砸了：

select NVL (skill.category, 'Total:') "Skill Category",
           count (training.code) "# of Trainings",
           count (project.code) "Projects Requiring Skill"
from skill join training on skill.code = training.code
           left join project on training.code = project.code
group by rollup (skill.category);

Skill Category  # of Trainings Projects Requiring Skill
--------------- -------------- ------------------------
Database                     4                        4
HR                           1                        0
Leadership                   1                        0
Printing                     3                        3
Sales                        3                        0
Web Design                  13                       13
Total:                      25                       20

我在这里做错了什么？我非常感谢任何帮助，对不起，如果我遗漏了一些明显的东西！

Answer 1

问题在于您加入了数据倍增。您有两个1-n关系，介于技能和培训之间以及培训和项目之间。结果是组合的倍增。

解决方案是分别进行两个摘要，然后使用join：

进行组合

select st.*, "projects req training"
from (select NVL(skill.category, 'Total:') "Skill Category",
             count(training.code) "# of Trainings"
      from skill left outer join
           training
           on skill.code = training.code
      group by rollup (skill.category)
     ) st join
     (select NVL(skill.category, 'Total:') "Skill Category",
             count(project.code) "projects req training"
      from skill left outer join
           project
           on skill.code = project.code
      group by rollup (skill.category)
     ) sp
     on st."Skill Category" = sp."Skill Category"

在这个版本中，我在子查询中使用left outer join，以确保每个子查询返回所有技能。这允许我在下一级使用inner join。我已将rollup留在子查询中，而不是在外部进行额外的聚合。

在甲骨文中统计 - 我做错了什么

1 个答案: