Using group by and searching with in a group

Question

I am looking for a way to group results in a table, then compare the results with in each group.
Here is some example data.
cid--eid--gid--seqid
101--m123--g123--111
100--m123--g123--111
102--m123--g123--111
200--m124--g124--211
201--m124--g124--211
202--m124--g124--210
300--m125--g223--311
301--m125--g223--311
302--m125--g223--311
303--m125--g223--311
400--m125--g223--411
401--m125--g223--411
402--m125--g223--411
403--m125--g223--411
403--m125--g223--412
If I group by eid, gid and order by cid I would get the following
Group 1:
101--m123--g123--111
100--m123--g123--111
102--m123--g123--111
Group 2:
200--m124--g124--211
201--m124--g124--211
202--m124-g124---210
Group 3:
300--m125--g223--311
301--m125--g223--311
302--m125--g223--311
303--m125--g223--311
Group 4:
400--m125--g223--411
401--m125--g223--411
402--m125--g223--411
403--m125--g223--411
403--m125--g223--412
Now I need to find out the cids in each group for which the seqid is NOT matching with majority in the group. For example, I need the output as
cid--eid--gid--seqid
202--m124--g124--210
403--m125--g223--412
What are the ways to do the above? One way I think is to use a temp table, but given the size of the table that may not be a possibility
Please let me know how to do this

Answer 1

with sample_table as (
                      select 101 cid,'m123' eid,'g123' gid,111 seqid from dual union all
                      select 100,'m123','g123',111 from dual union all
                      select 102,'m123','g123',111 from dual union all
                      select 200,'m124','g124',211 from dual union all
                      select 201,'m124','g124',211 from dual union all
                      select 202,'m124','g124',210 from dual union all
                      select 300,'m125','g223',311 from dual union all
                      select 301,'m125','g223',311 from dual union all
                      select 302,'m125','g223',311 from dual union all
                      select 303,'m125','g223',311 from dual union all
                      select 400,'m125','g223',411 from dual union all
                      select 401,'m125','g223',411 from dual union all
                      select 402,'m125','g223',411 from dual union all
                      select 403,'m125','g223',412 from dual
select  cid,
        eid,
        gid,
        seqid
  from  (
         select  cid,
                 eid,
                 gid,
                 seqid,
                 seqid_cnt,
                 max(seqid_cnt) over(partition by grp) max_cnt
           from  (
                  select  cid,
                          eid,
                          gid,
                          seqid,
                          grp,
                          count(*) over(partition by grp,seqid) seqid_cnt
                    from  (
                           select  s.*,
                                   cid - row_number() over(partition by eid,gid order by cid) grp
                             from  sample_table s
  where seqid_cnt != max_cnt
       CID EID  GID       SEQID
       202 m124 g124        210
       403 m125 g223        412
SQL> SY.