Oracle sql - select distinct columns based on few columns - sql

Question

I have a select query like,
select a, b, c, d, e, f, g, h, i, j from sample_table
I need to have distinct set of records from this table, so I put
select distinct a, b, c, d, e, f, g, h, i, j from sample_table
But still, the duplicate columns are coming in the result set, as i,j is differing with a minor change like result, result1, RESULT. I need to get rid of this minor change but want to have it in the result set.
How do I select distinct columns of a,b,c,d,e,f,g,h and also have i,j in the result set.

Answer 1

You can do this using the analytic functions:
select a, b, c, d, e, f, g, h, i, j
from (select st.*,
row_number() over (partition by a, b, c, d, e, f, g, h order by a) as seqnum
from sample_table
) st
where seqnum = 1;
This ensures that the values of i and j come from the same row.

Answer 2

SELECT DISTINCT removes duplicate rows.
If you consider certain values to be "the same" either within a column or between two columns, then before rows containing them can be seen as duplicates by the DBMS you have to make them actually the same.
Within a column you can convert each possible variation to one particular variation. This is called converting to a canonical or normal form.
select distinct ...,
case i when "result1" then "result"
else "RESULT" then "result"
else "result" then "result"
else "dOg" then "dog"
...
end as i,
convert_to_upper_case(j) as j,
correct_spelling(k) as k
from sample_table
If you want to consider values to be the same across columns then you can convert them in that way and compare the canonical forms. Or you can write an expression that compares them and output a single value both columns. This is called an equivalence relation.
select distinct ...,g,h,i, i as j
from sample_table
where ...
AND my_canonical_form(g) = my_canonical_form(h)
AND equivalent_according_to_me(i,j)
That can be used in generating sample_table if j wasn't really supposed to be different from i there:
select distinct ..., t.i, t.i as j -- no u.j
from t,u where ... and close_enough(t.i,u.j)
The idea is that canonical_form(x) = canonical_form(y) exactly when equivalent(x,y).
You can either keep both i and j columns or drop one if you want.

Answer 3

May be you can try:
select distinct a, b, c, d, e, f, g, h, min(upper(i)) i, min(upper(j)) j from sample_table
group by a, b, c, d, e, f, g, h;
You can consider using min or max combined with substring, upper, lower or whichever suits your requirement.
As alex poole has pointed out, you can also consider having a column with timestamp, so that the latest or the earliest record can be displayed in the result-set.

Answer 4

In addition to Nishanthi Grashia answer: if you have to show all the values of the different columns, you can use listagg as aggregate function:
select same1,same2,same3,same4,same5,same6,same7,same8,
listagg(diff1, ',') within group (order by 1,2,3,4,5,6,7,8)
, listagg(diff2, ',') within group (order by 1,2,3,4,5,6,7,8) from (
select 1 as same1, 2 as same2 ,3 as same3, 4 as same4,5 as same5,6 as same6,7 as same7, 8 as same8,'m' as diff1,'n' as diff2 from dual
union
select 1,2,3,4,5,6,7,8,'n','o' from dual
union
select 2,3,4,5,6,7,8,9,'p','q' from dual
union
select 2,3,4,5,6,7,8,9,'p','x' from dual
) qry1 group by same1,same2,same3,same4,same5,same6,same7,same8