In SQL we need to transform a table in the following way:
Table1:
+-----+---------+-----------+
| ID | insured | DOD |
+-----+---------+-----------+
| 123 | Pam | 6/18/2013 |
| 123 | Nam | 2/12/2010 |
| 123 | Tam | 2/10/2013 |
| 456 | Jessi | 4/6/2003 |
| 457 | Ron | 4/10/2010 |
| 457 | Tom | 5/5/2008 |
+-----+---------+-----------+
Desired output table:
+-----+---------+-----------+-----------+-----------+
| ID | insured | DOD1 | DOD2 | DOD3 |
+-----+---------+-----------+-----------+-----------+
| 123 | Pam | 6/18/2013 | 2/12/2010 | 2/10/2013 |
| 456 | Jessi | 4/6/2003 | null | null |
| 457 | Ron | 4/10/2010 | 5/5/2008 | null |
+-----+---------+-----------+-----------+-----------+
I have seen somewhere that we can use pivot and unpivot, but I am not sure how can I use it here.
Your help is much appreciated.
Assuming that you really want -- or can accept -- the dates in descending order, then you can use conditional aggregation for this:
select id,
max(case when seqnum = 1 then insured end) as insured,
max(case when seqnum = 1 then dod end) as dod_1,
max(case when seqnum = 2 then dod end) as dod_2,
max(case when seqnum = 3 then dod end) as dod_3
from (select t.*,
row_number() over (partition by id order by dod desc) as seqnum
from t
) t
group by id;
If you want to preserve the original ordering, then your question does not have enough information. If you have a column with the ordering, that can be used for row_number().