SQL子查询返回排名2

Question

I have a question about writing a sub-query in Microsoft T-SQL. From the original table I need to return the name of the person with the second most pets. I am able to write a query that returns the number of perts per person, but I'm not sure how to write a subquery to return rank #2.

Original table:

+—————————-——+———-————-+
|   Name     |  Pet    | 
+————————————+————-————+
| Kathy      |  dog    | 
| Kathy      |  cat    |
| Nick       |  gerbil | 
| Bob        |  turtle | 
| Bob        |  cat    | 
| Bob        |  snake  | 
+—————————-——+—————-———+

I have the following query:

SELECT Name, COUNT(Pet) AS NumPets
FROM PetTable
GROUP BY Name
ORDER BY NumPets DESC

Which returns:

+—————————-——+———-————-+
|   Name     | NumPets | 
+————————————+————-————+
| Bob        |  3      | 
| Kathy      |  2      | 
| Nick       |  1      | 
+—————————-——+—————-———+

Answer 1

The ANSI standard method is:

OFFSET 1 FETCH FIRST 1 ROW ONLY

However, most databases have their own syntax for this, using limit , top or rownum . You don't specify the database, so I'm sticking with the standard.

Answer 2

You are using TSQL So:

WITH C AS (
    SELECT  COUNT(Pet) OVER (PARTITION BY Name) cnt
            ,Name
    FROM PetTable
)
SELECT TOP 1 Name, cnt AS NumPets
FROM C
WHERE cnt = 2

Answer 3

This is how you could use ROW_NUMBER to get the result.

SELECT *
FROM(
SELECT ROW_NUMBER() OVER (ORDER BY COUNT(name) DESC) as RN, Name, COUNT(NAME) AS COUNT
FROM PetTable
GROUP BY Name
) T
WHERE T.RN = 2

Answer 4

In MSSQL you can do this:

SELECT PetCounts.Name, PetCounts.NumPets FROM (
  SELECT
    RANK() OVER (ORDER BY COUNT(Pet) DESC) AS rank,
    Name, COUNT(Pet)as NumPets
  FROM PetTable
  GROUP BY Name
) AS PetCounts
WHERE rank  = 2

This will return multiple rows if they have the same rank. If you want to return just one row you can replace RANK() with ROW_NUMBER()