选择COUNT个值，按同一列分组，在另一列上不同

Question

Here's a sample schema:

| type_id | university_id |
---------------------------
|    1    |      u01      |
|    2    |      u01      |
|    2    |      u01      |
|    3    |      u02      |
|    4    |      u02      |

And what I want is a count of the amount of records that have the same university_id , the university_id itself, and it must not include duplicates of type_id , so the result would look like this for the same data given above:

| university_id | COUNT(university_id) |
----------------------------------------
|      u01      |          2           |
|      u02      |          2           |

Right now I'm trying the following, but it will clearly give me a COUNT(u01) of 3 instead of 2

SELECT university_id, COUNT(university_id) FROM table GROUP BY university_id

Is it possible to do what I want, and if so how?

Answer 1

You should count by type_id uniquely not by university_id

SELECT university_id, COUNT(DISTINCT type_id) TotalCount
FROM   tableName
GROUP  BY university_id

• SQLFiddle Demo

Answer 2

You need a count(distinct) on the type_id column:

select university_id, count(distinct type_id)
from table
group by university_id