SQL get array in postgresql

Question

Have a table(test):

  color              |  count_var  | 
 --------------------+-------------+
  red & white        |          15 |
  black & white      |          10 |
  black              |          10 |
  black & white      |          15 |
  red & white        |          15 |              

1. At first I need to GROUP BY the 'count_var'
2. Then I need to select the max value and corresponding 'count_var'

    SELECT MAX(mycount) as mode
    FROM (SELECT COUNT(*) as mycount, count_var
          FROM test
          GROUP BY count_var) as t;

Have the result

 mode 
------
 3

What I need at first? The result like

 mode | count_var
------+----------
 3    |     15

Then I need to insert this result at another table to the field with type integer[].

The inserting result should be {3,15}

How can I do this?

Answer 1

Well, you can do:

SELECT ARRAY[num_mycount, mycount]
FROM (SELECT count_var, COUNT(*) as mycount,
             COUNT(*) OVER (PARTITION BY COUNT(*)) as num_mycount
      FROM test
      GROUP BY count_var
     ) t
ORDER BY mycount DESC
FETCH FIRST 1 ROW ONLY;

Amusingly, I don't think I have ever used COUNT(*) OVER (PARTITION BY COUNT(*)) before. I've gotten used to mixing window functions with aggregations, but I can imagine some people will find this a little confusing.

The first COUNT(*) is a window function counting the number of rows in a partition of the result set -- that is, after the aggregation. The second is the aggregation COUNT(*) . So the effect is to count the number of values that have the same count after the aggregation.