Getting the third moving date - sql

Question

I need to get the third moving date based on address for each person. Either status change or address change will make DT change. For example, for PERSON_ID:1, the third moving date should be 07/16/2016. Thanks!
The data is as below:
PERSON_ID STATUS DT ADDRESS
1 12 5/6/2016 3
1 6 5/8/2016 3
1 7 6/5/2016 3
1 1 6/13/2016 3
1 12 6/20/2016 1
1 17 7/8/2016 1
1 1 7/11/2016 1
1 12 7/16/2016 2
1 3 12/6/2016 2
2 5 3/11/2016 5
2 1 5/15/2016 4
2 6 7/18/2016 6
2 12 7/21/2016 6

Answer 1

Using row_number() and group by for the min(dt) per address:
Note: This will not work correctly if the person moves between the same addresses.
select
Person_id
, dt = convert(char(10),dt,120)
, Address
from (
select
person_id
, dt = min(dt)
, address
, rn = row_number() over (partition by person_id order by min(dt))
from t
group by person_id, address
) s
where rn = 3
rextester demo: http://rextester.com/VLTUU16478
returns:
+-----------+------------+---------+
| Person_id | dt | Address |
+-----------+------------+---------+
| 1 | 2016-07-16 | 2 |
| 2 | 2016-07-18 | 6 |
+-----------+------------+---------+
To solve this correctly for a person moving between the same addresses, you have to address the gaps and islands problem.
Adding an additional subquery to the above solution so we can identify and group by the islands:
select
Person_id
, dt = convert(char(10),dt,120)
, Address
from (
select
person_id
, dt = min(dt)
, address
, rn = row_number() over (partition by person_id order by min(dt))
from (
select
person_id
, address
, dt
, island = row_number() over (partition by person_id order by dt)
- row_number() over (partition by person_id, address order by dt)
from t
) s
group by person_id, address, island
) s
where rn = 3
rextester demo: http://rextester.com/PPIH49666
returns:
+-----------+------------+---------+
| Person_id | dt | Address |
+-----------+------------+---------+
| 1 | 2016-07-16 | 3 |
| 2 | 2016-07-18 | 5 |
+-----------+------------+---------+

Answer 2

I'm not sure i understood what you need to get but I think what you're trying to do is something like:
SELECT * FROM
(
SELECT person_id,adress,[status],DT, rank() over (partition by person_id order by adress,dt)-1 as movement
FROM #t
WHERE person_id=1
) t
WHERE t.movement=3

Answer 3

Question is not entirely clear
select PERSON_ID, STATUS, DT, ADDRESS
from ( select PERSON_ID, STATUS, DT, ADDRESS
, row_number() over (partition by person order by dt) as rn
from ( select PERSON_ID, STATUS, DT, ADDRESS
, row_number() over ( partition by PERSON_ID, address order by dt) as rn
from table
) tt
where tt.rn = 1
) rr
where rr.rn = 3