Cummulative SUM based on columns - sql

Question

I have a table with values like this:
I want to get cumulative sum based on the ID and year, so it should return an output like this i.e for id- 1 and year 2010 the sum of records will be 2.
id-2 and year 2010 the sum of records will be 1 and
id- 2 and for year 2011 it will be 1+1 = 2 i.e i require a running total for each id in ascending order based upon year.
similarly for id =3 Sum will be 1 , for id 4 will be 1 based on the year. for 5 it will be 3 for yr 2014 , for 2015 it will be sum of count previous yr + sum of count current yr i.e it will be 3 + 1 = 4 and for year 2016 it will be 3+ 1+1 = 5. Hence what is to be done. Could someone please help?

Answer 1

No need to make thinks more complicated than they need to be...
IF OBJECT_ID('tempdb..#TestData', 'U') IS NOT NULL
DROP TABLE #TestData;
CREATE TABLE #TestData (
ID INT NOT NULL,
[Year] INT NOT NULL
);
INSERT #TestData (ID, Year) VALUES
(1, 2010), (1, 2010), (2, 2010), (2, 2011),
(3, 2012), (4, 2013), (5, 2014), (5, 2014),
(5, 2014), (5, 2015), (5, 2016);
--=======================================
SELECT
tdg.ID,
tdg.Year,
RunningCount = SUM(tdg.Cnt) OVER (PARTITION BY tdg.ID ORDER BY tdg.Year ROWS UNBOUNDED PRECEDING)
FROM (
SELECT td.ID, td.Year, Cnt = COUNT(1)
FROM #TestData td
GROUP BY td.ID, td.Year
) tdg;
Results...
ID Year RunningCount
----------- ----------- ------------
1 2010 2
2 2010 1
2 2011 2
3 2012 1
4 2013 1
5 2014 3
5 2015 4
5 2016 5

Answer 2

this is more nesting than I would like, and I feel there is a better way to do this with maybe only one windows function but I can't get past not having a unique row for your data.
SELECT id,
year ,sum(c) OVER (
PARTITION BY id ORDER BY year rows unbounded preceding
)
FROM (
SELECT id,
year,
count(rn) c
FROM (
SELECT id,
year,
row_number() OVER (
ORDER BY year
) AS rn
FROM your_table -- you will need to change this to your table
) a
GROUP BY id,
year
) a
what we do is first build the data with a row number so now everything is unique, after that we then do a count on that unique row number and do windows function to do a running total for the count of rows by year.

Answer 3

There are many ways to do this. Here is one of them, with an inner query:
create table #table_name
(
UserID int,
Year int
)
INSERT INTO #table_name (UserID, Year)
VALUES
(1, 2010)
,(1,2010)
,(2,2010)
,(2,2011)
,(3,2012)
,(4,2013)
,(5,2014)
,(5,2014)
,(5,2014)
,(5,2015)
,(5,2016)
SELECT
UserID
,YEAR
,(SELECT COUNT(Year) FROM #table_name WHERE Year <= tt.Year AND UserID = tt.UserID)
FROM
#table_name AS tt
GROUP BY UserID, Year
you can also use row number over (edit : see below answer for this technique, I think it is a little bit too complicated for such a simple task). The query above returns your required output
+--------+------+-------+
| UserID | Year | COUNT |
+--------+------+-------+
| 1 | 2010 | 2 |
| 2 | 2010 | 1 |
| 2 | 2011 | 2 |
| 3 | 2012 | 1 |
| 4 | 2013 | 1 |
| 5 | 2014 | 3 |
| 5 | 2015 | 4 |
| 5 | 2016 | 5 |
+--------+------+-------+