Query to find out two counts on each column

Question

Hi
can you help for the below query.
I wanted to find out two count values on each column, ie above or equal 5 count and below 5 count.
Table:
Q1 Q2 Q3 (Q1, Q2, Q3 are columns in a table.)
1 3 5
6 7 4
8 5 1
2 8 5
Query should display, like below.
Expected Result:
colname >=5 count <5 count
Q1 2 2
Q1 3 1
Q3 2 2
Thanks in advance.

Answer 1

The query itself can be simplified as follows:
SQL> with t as (
  2       select level lvl from dual
  3       connect by level < 4)
  4  select colname
  5       , sum(case when val >= 5 then 1 else 0 end) "count >= 5"
  6       , sum(case when val < 5 then 1 else 0 end) "count < 5"
  7  from (
  8       select decode (lvl, 1, 'Q1', 2, 'Q2', 'Q3') colname
  9               , decode (lvl, 1, q1, 2, q2, q3) val
10               , lvl
11       from (select q1, q2, q3
12                       from q),
13                       t)
14  group by colname, lvl
15  order by lvl;Lines 8-10 is what pivots the cartesian join (CJ) of q & t. It helps to explain it by breaking down each part.
The CJ produces (I added the lvl into the output below for explanation purposes):
SQL> with t as (
  2       select level lvl from dual
  3       connect by level < 4)
  4  select q1, q2, q3, lvl
  5  from q,
  6  t;
        Q1         Q2         Q3        LVL
         1          3          5          1  <-- line a1
         6          7          4          1  <-- line a2
         8          5          1          1
         2          8          5          1
         1          3          5          2  <-- line a5
         6          7          4          2
         8          5          1          2
         2          8          5          2
         1          3          5          3
         6          7          4          3
         8          5          1          3
         2          8          5          3
12 rows selected.and the CJ from above now allows us to pivot the results using the LVL column as the "decoding" key.
SQL> with t as (
  2       select level lvl from dual
  3       connect by level < 4)
  4  select decode (lvl, 1, 'Q1', 2, 'Q2', 'Q3') colname
  5       , decode (lvl, 1, q1, 2, q2, q3) val
  6       , lvl
  7  from (select q1, q2, q3
  8         from q),
  9         t;
COLNAME           VAL        LVL
Q1                  1          1 <-- line b1
Q1                  6          1 <-- line b2
Q1                  8          1
Q1                  2          1
Q2                  3          2 <-- line b5
Q2                  7          2
Q2                  5          2
Q2                  8          2
Q3                  5          3
Q3                  4          3
Q3                  1          3
Q3                  5          3
12 rows selected.Line b1:
The decodes hit line a1 of the CJ and since lvl = 1, we get colname=Q1 and the corresponding Q1 row value (val=1) of line a1 and finally the value of 1 for lvl.
Line b2:
For decode of line a2 of the CJ, lvl = 1 again and we get colname=Q1 and the corresponding Q1 row value (val=6) of line a2 and finally the value of 1 for lvl.
Line b5:
For decode of line a5 of the CJ, lvl = 2 and we get colname=Q2 and the corresponding Q2 row value (val=3) of line a5 and finally the value of 2 for lvl.
The other rows of the pivoted result set above are obtained similarly and from this result we can now perform the aggregation for the final result.
The WITH query is a called subquery factoring. It allows you to reference the subquery multiple places in the query.
HTH