SQL: Update Nulls with Previous Timeframe Values - sql-server

Question

I currently have a resulting table in SQL that shows during which time period (college semester) a person's address changed This doesn't happen every time period, so some rows are showing null as expected I am needing these to update ("fill down") for each subsequent time period until a new address change is entered
I included these two IDs because they represent the two possible cases of what I am seeing:
-ID 1234 should fill the preceding Terms with the Sequence 1 county (shown here)
-ID 5678 should fill the preceding Terms with the Sequence 1 county as well (CLAY in this case) based on a previously joined table
Currently, I am showing something along the lines of:
ID TERM COUNTY SEQUENCE
------------------------------------------
1234 201308 null null
1234 201401 null null
1234 201408 ORANGE 1
1234 201501 null null
1234 201505 null null
1234 201508 OSCEOLA 3
1234 201601 null null
5678 201301 null null
5678 201305 null null
5678 201308 ST JOHNS 3
5678 201401 null null
5678 201405 null null
5678 201408 null null
5678 201501 null null
5678 201505 DUVAL 4
And I need the output to look like:
ID TERM COUNTY SEQUENCE
---------------------------------------------
1234 201308 ORANGE null
1234 201401 ORANGE null
1234 201408 ORANGE 1
1234 201501 ORANGE null
1234 201505 ORANGE null
1234 201508 OSCEOLA 3
1234 201601 OSCEOLA null
5678 201301 CLAY null
5678 201305 CLAY null
5678 201308 ST JOHNS 3
5678 201401 ST JOHNS null
5678 201405 ST JOHNS null
5678 201408 ST JOHNS null
5678 201501 ST JOHNS null
5678 201505 DUVAL 4
This is my first time coming across an update clause need like this, so any insight you may be able to provide will be greatly appreciated!
*I am not sure how much of the previous code will be relevant, but here is essentially the temp table code that feeds into the final output ("PIDM" is the ID):
DROP TABLE #ADDRESS_PT_1--, #ADDRESS_PT_2
GO
SELECT SPRADDR_PIDM 'PIDM', Y.TERM, SPRADDR_SEQNO 'SEQNO', SPRADDR_STAT_CODE 'STATE', SPRADDR_CNTY_CODE 'CNTY',
BANNR_TERM = CASE
WHEN Y2.TERM IS NULL THEN Y.TERM
ELSE Y2.TERM
END
INTO #ADDRESS_PT_1
FROM SPRADDR
LEFT OUTER JOIN RCACYR Y2
ON (SPRADDR_ACTIVITY_DATE BETWEEN Y2.BEGIN_DATE AND Y2.END_DATE
AND SUBSTRING(Y2.TERM,5,2) IN ('50','80','10')),
RCACYR Y
WHERE SPRADDR_ACTIVITY_DATE BETWEEN Y.BEGIN_DATE AND Y.END_DATE
AND SUBSTRING(Y.TERM,5,2) IN ('05','08','01')
AND SPRADDR_ATYP_CODE = 'MA'
ORDER BY SPRADDR_PIDM, SPRADDR_SEQNO
GO
/* Get the individuals addresses for each term */
SELECT *
--INTO #ADDRESS_PT_2
FROM #ADDRESS_PT_1 X
LEFT JOIN RCCNTY C
ON C.COUNTY = X.CNTY
WHERE X.SEQNO = (SELECT MAX(A.SEQNO)
FROM #ADDRESS_PT_1 A
WHERE X.PIDM = A.PIDM
AND X.TERM = A.TERM)
--AND X.PIDM = 5678
ORDER BY X.PIDM, X.SEQNO
GO
The output from this is:
PIDM TERM SEQNO STATE CNTY BANNR_TERM COUNTY COUNTY_TITLE COUNTY_REGION COUNTY_REGION_TITLE
5678 201108 1 FL CLAY 201108 CLAY CLAY 2 Northeast Florida
5678 201308 3 FL ST J 201308 ST J ST. JOHNS 2 Northeast Florida
5678 201505 5 FL DUVA 201550 DUVA DUVAL 2 Duval County

Answer 1

I put in CTE the sample you provided. Then I OUTER APPLY (p) previous row with NOT NULL COUNTY, and another OUTER APPLY that gets the row with [SEQUENCE] = 1 for each ID. Instead of FROM cte in last OUTER APPLY use table (FROM SPRADDR) that have rows with [SEQUENCE] = 1 which might not be in CTE.
;WITH cte AS (
SELECT *
FROM (VALUES
(1234, 201308, null, null),
(1234, 201401, null, null),
(1234, 201408, 'ORANGE', 1),
(1234, 201501, null, null),
(1234, 201505, null, null),
(1234, 201508, 'OSCEOLA', 3),
(1234, 201601, null, null),
(5678, 201301, null, null),
(5678, 201305, null, null),
(5678, 201308, 'ST JOHNS', 3),
(5678, 201401, null, null),
(5678, 201405, null, null),
(5678, 201408, null, null),
(5678, 201501, null, null),
(5678, 201505, 'DUVAL', 4)
) as t(ID, TERM, COUNTY, [SEQUENCE])
)
SELECT c.ID,
c.TERM,
COALESCE(c.COUNTY,p.COUNTY,p1.COUNTY) as COUNTY,
c.[SEQUENCE]
FROM cte c
OUTER APPLY (
SELECT TOP 1 COUNTY
FROM cte
WHERE ID = c.ID
AND TERM < c.TERM
AND COUNTY IS NOT NULL
ORDER BY TERM DESC) as p
OUTER APPLY (
SELECT TOP 1 COUNTY
FROM cte
WHERE ID = c.ID
AND [SEQUENCE] = 1
ORDER BY TERM DESC) as p1
Will give you:
ID TERM COUNTY SEQUENCE
1234 201308 ORANGE NULL
1234 201401 ORANGE NULL
1234 201408 ORANGE 1
1234 201501 ORANGE NULL
1234 201505 ORANGE NULL
1234 201508 OSCEOLA 3
1234 201601 OSCEOLA NULL
5678 201301 NULL NULL
5678 201305 NULL NULL
5678 201308 ST JOHNS 3
5678 201401 ST JOHNS NULL
5678 201405 ST JOHNS NULL
5678 201408 ST JOHNS NULL
5678 201501 ST JOHNS NULL
5678 201505 DUVAL 4

Answer 2

See an example:
SELECT * into tbl_filltest FROM (
VALUES (1,Null),(2,Null),(3,5),(4,Null),(5,Null),(6,Null),(7,4),(8,Null),9,Null),(10,1)
) as t(c1,c2)
GO
SELECT * FROM tbl_filltest
GO
;WITH GoodValues as (SELECT * FROM tbl_filltest WHERE c2 is not null),
NullValues as (SELECT * FROM tbl_filltest WHERE c2 is null)
UPDATE n SET c2 = g1.c2 FROM GoodValues as g1
OUTER APPLY (SELECT MAX(c1) as Min_c1 FROM GoodValues as i WHERE g1.c1 > i.c1) as g2
INNER JOIN NullValues as n
ON n.c1 > IsNull(g2.Min_c1,0) and n.c1 < g1.c1
GO
SELECT * FROM tbl_filltest
GO