计算一些特殊情况的查询

Question

我有一个table

考试（studentID、subjectCode、examDate、examScore）

还有很多值，例如

INSERT [dbo].[Exam] ([studentID], [subjectCode], [examDate], [examScore]) VALUES (N'ST01', N'PRJ', CAST(0x253C0B00 AS Date), 5)
INSERT [dbo].[Exam] ([studentID], [subjectCode], [examDate], [examScore]) VALUES (N'ST01', N'WIG', CAST(0x253C0B00 AS Date), 1)
 . . .
INSERT [dbo].[Exam] ([studentID], [subjectCode], [examDate], [examScore]) VALUES (N'ST09', N'WIG', CAST(0x253C0B00 AS Date), 4)
INSERT [dbo].[Exam] ([studentID], [subjectCode], [examDate], [examScore]) VALUES (N'ST09', N'WIG', CAST(0x263C0B00 AS Date), 9)
INSERT [dbo].[Exam] ([studentID], [subjectCode], [examDate], [examScore]) VALUES (N'ST10', N'MAD', CAST(0x253C0B00 AS Date), 3)
INSERT [dbo].[Exam] ([studentID], [subjectCode], [examDate], [examScore]) VALUES (N'ST10', N'MAE', CAST(0x253C0B00 AS Date), 4)

我必须计算学生的所有平均分数，以便：（抱歉格式太糟糕）

s_i is the average score of the student i^th with i=1…M where M is the total number of students N is the total number of subjects. The average score of the student s_i can be calculated by:

Where x_(i,k) is the last exam’s score of the student s_i on the subjectk^th. The last exam’s score means: student may retake the exam, and only the final result will be involved into the average score calculation. The x_(i,k) may receive Null value if student s_i did not take exam on subject k^th.

我只能做类似的事情

select studentID, sum(examScore)/count(subjectCode) from Exam group by studentID

下半年如何处理这些特殊情况的问题？

Answer 1

据我了解你的问题，你想要每个学生的平均值，只考虑每个科目的最新分数。如果是，可以使用window函数进行过滤，然后聚合：

select studentid, avg(examscore) avgscore
from (
    select e.*, 
        row_number() over(partition by studentid, subjectcode order by examdate desc) rn
    from exam e
) e
where rn = 1
group by studentid

如果要将“缺失”科目计为0，那就有点复杂了。您通常会 cross join 不同学生和科目的列表，然后将 table 与 left join:

select st.studentid, avg(coalesce(e.examscore, 0)) avgscore
from (select distinct studentid from exam) st
cross join (select distinct subjectcode from exam) su
left join (
    select e.*, 
        row_number() over(partition by studentid, subjectcode order by examdate desc) rn
    from exam e
) e 
    on  e.studentid   = st.studentid 
    and e.subjectcode = su.subjectcode 
    and e.rn = 1
group by st.studentid

在现实生活中，您可能有单独的引用 table 来存储学生和科目，您将使用它们来代替 select distinct_ 子查询。

计算一些特殊情况的查询

Calculate query with some special case

sql

sql-server

average

aggregate-functions

greatest-n-per-group