如何在SQL中获得MAX（COUNT）？

Question

I would like to get the name of the entity which has MAX count in the dataset and would also like to get the count of the same.

SELECT 
    STORE_TYPE AS 'FREQUENT CHANNEL', 
    COUNT(STORE_TYPE) AS 'TOTAL TRANSACTIONS'
FROM 
    TRANSACTION_INFO
GROUP BY 
    STORE_TYPE

This returns the count of the STORE_TYPE along with the TOTAL_TRANSACTIONS .

Now, I would like to fetch the STORE_TYPE along with TOTAL TRANSACTIONS which has performed the best (MAX).

Thanks in advance!

Answer 1

If you want one row, the logic would be group by to calculate the count. Then order by and some way of limiting the result to the top:

SELECT STORE_TYPE, COUNT(*) AS cnt
FROM TRANSACTION_INFO
GROUP BY STORE_TYPE
ORDER BY cnt DESC
FETCH FIRST 1 ROW ONLY;

In the event of ties, this only returns one arbitrary row with the highest value. The best way to handle ties depends on the database.

FETCH FIRST 1 ROW ONLY is the ISO/ANSI standard for limiting the result set to one row. Your database may use a different syntax for this.

I also strongly discourage you from using single quotes for column aliases. Use names that don't need to be escaped -- that is the best advise. Otherwise, use the escape mechanism for your database, typically double quotes, backticks, or square braces.

EDIT:

The solution in SQL Server that includes ties is:

SELECT TOP (1) STORE_TYPE, COUNT(*) AS cnt
FROM TRANSACTION_INFO
GROUP BY STORE_TYPE
ORDER BY cnt DESC;

If you only want one row, remove the WITH TIES .

Answer 2

You can simply ORDER BY the max and then limit to a single row.

SELECT 
STORE_TYPE AS 'FREQUENT CHANNEL', 
COUNT(STORE_TYPE) AS 'TOTAL TRANSACTIONS'
FROM TRANSACTION_INFO
GROUP BY STORE_TYPE
ORDER BY 'TOTAL TRANSACTIONS' DESC
LIMIT 1

The way to LIMIT depends on the SQL dialect.

Answer 3

The other answers work well, as long as there's a single store_type that is doing best. If you also want to account for multiple ones tied in first place you should use:

with s (t, c) as (
  select store_type, count(*)
    from transaction_info group by store_type
)
select
    t as 'FREQUENT CHANNEL',
    c as 'TOTAL TRANSACTIONS'
  from s
  where c = (select max(c) from s)

This query will show you one or more rows.