Special Aggregation

Question

I have the following record set
1 s1 1/13/2006
1 s2 1/13/2006
1 s1 1/14/2006
1 s2 1/14/2006
1 s1 1/15/2006
1 s2 1/15/2006
1 s3 1/16/2006
1 s3 1/17/2006
1 s3 1/18/2006
1 s4 1/19/2006
1 s4 1/20/2006
1 s4 1/21/2006
1 s4 1/23/2006
1 s4 1/24/2006
1 s4 1/27/2006
1 s4 1/28/2006
2 s1 1/13/2006
2 s2 1/13/2006
2 s1 1/14/2006
2 s2 1/14/2006
2 s1 1/15/2006
2 s2 1/15/2006
2 s3 1/16/2006
2 s3 1/17/2006
2 s3 1/18/2006
2 s4 1/19/2006
2 s4 1/20/2006
2 s4 1/21/2006
2 s4 1/23/2006
2 s4 1/24/2006
2 s4 1/27/2006
2 s4 1/28/2006
I need to aggregate those records as the following
1 s1,s2 1/13/2006 1/15/2006
1 s3 1/16/2006 1/18/2006
1 s4 1/19/2006 1/21/2006
1 s4 1/23/2006 1/24/2006
1 s4 1/27/2006 1/28/2006
2 s1,s2 1/13/2006 1/15/2006
2 s3 1/16/2006 1/18/2006
2 s4 1/19/2006 1/21/2006
2 s4 1/23/2006 1/24/2006
2 s4 1/27/2006 1/28/2006
Which means that I have to group them first according to te first column (1, 2), and then according to the second one but according to the consecutive days! From the example above, you may notice that if the second column repeated for the same group (first column) in the same day, the output requires concatenating them (s1, s2), and this pattern lasts from 1/13/2006 till 1/15/2006
If break in these days is there then I need a separate group; for instance, 2 with s4 has been split into three groups; from 1/19/2006 till 1/21/2006, from 1/23/2006 till 1/24/2006, and from 1/27/2006 till 1/28/2006, and this is only because the break in the day sequence!
Can anyone help please?
Thanks in advance

Answer 1

23:04:21 session_152> select n,scbp,from_dt, thru_dt
23:04:30   2  from (
23:04:30   3  select n,scbp,mg,mlg, min(d) from_dt, max(d) thru_dt
23:04:30   4  from (
23:04:30   5  select n,d,scbp,mg,max(lg) over (partition by n,mg order by d) mlg
23:04:30   6  from (
23:04:30   7  select n,d,scbp,mg
23:04:30   8        ,case when trunc(lag(d) over (partition by n,mg order by d)) = trunc(d)-1
23:04:30   9                then null
23:04:30  10              else
23:04:30  11                 row_number() over (partition by n,mg order by d)
23:04:30  12              end lg
23:04:30  13  from (
23:04:30  14  select n,d,scbp,max(g) over (partition by n order by d) mg
23:04:30  15  from (
23:04:30  16  select n,d,scbp
23:04:30  17        ,case when lag(scbp) over (partition by n order by d) = scbp
23:04:30  18                then null
23:04:30  19              else
23:04:30  20                row_number() over (partition by n order by d)
23:04:30  21         end g
23:04:30  22  from (
23:04:30  23  select n, d, max(ltrim(sys_connect_by_path(v,','),',')) scbp
23:04:30  24  from (
23:04:30  25  select n,v,d
23:04:30  26        ,row_number() over (partition by n, d order by v) c
23:04:30  27        ,row_number() over (partition by n, d order by v) - 1 p
23:04:30  28  from t
23:04:30  29  )
23:04:30  30  start with c = 1
23:04:30  31  connect by prior c = p and prior n = n and prior d = d
23:04:30  32  group by n,d
23:04:30  33  ) ) ) ) )
23:04:30  34  group by n,scbp,mg,mlg
23:04:30  35  )
23:04:30  36  ;
         N SCBP                 FROM_DT    THRU_DT
         1 s1,s2                01/13/2006 01/15/2006
         1 s3                   01/16/2006 01/18/2006
         1 s4                   01/19/2006 01/21/2006
         1 s4                   01/23/2006 01/24/2006
         1 s4                   01/27/2006 01/28/2006
         2 s1,s2                01/13/2006 01/15/2006
         2 s3                   01/16/2006 01/18/2006
         2 s4                   01/19/2006 01/21/2006
         2 s4                   01/23/2006 01/24/2006
         2 s4                   01/27/2006 01/28/2006
10 rows selected.