How to loop through dates, 30 times to find historical information at points in time? - sql

Question

I am trying to find out the transaction details of customers at different points in time in the past. I came up with a query but I have to change the date every single time in my declare statement. Is there a way to loop through the dates to get data back x amount of days in the past?
DECLARE #StartDate AS Date
SET #StartDate = '2018-10-30'
SELECT u.member_id AS Member_id
,CAST(MAX(pe.executed_time) AS Date) AS Max_Date
,DATEDIFF(dd,#StartDate,MAX(pe.executed_time))*-1+1 AS R
,COUNT(*) AS F
,SUM(pi.Price) AS M
FROM [user] AS u
LEFT JOIN purchase_entry AS pe
ON u.member_id=pe.member_id
LEFT JOIN purchase_item AS pi
ON pe.order_id=pi.order_id
WHERE 1=1
AND pe.[status] = 'purchase'
AND pe.executed_time <= #StartDate -- Change the declared Date to have historical information.
GROUP BY u.id, u.member_id
ORDER BY Recency
I need a way to have my query to loop through the #StartDate for 30 days in the past for instance. For ex: #StartDate = '2018-11-30' then #StartDate '2018-11-29' , and so on ...

Answer 1

You can list the dates that you want:
WITH dates as (
SELECT v.*
FROM (VALUES (CONVERT(DATE, '2018-11-30')), (CONVERT(DATE, '2018-11-29'))
) v(dte)
)
SELECT dates.dte, u.member_id, . . .
FROM dates CROSS JOIN
[user] u JOIN
purchase_entry pe
ON u.member_id = pe.member_id LEFT JOIN
purchase_item pi
ON pe.order_id = pi.order_id
WHERE pe.[status] = 'purchase' AND
pe.executed_time <= date.dte -- Change the declared Date to have historical information.
GROUP BY v.dte, u.id, u.member_id
ORDER BY Recency;
For a specific period of dates, you can construct them using a recursive CTE:
with dates as (
select convert(date, '2018-11-30') as dte, 1 as n
union all
select dateadd(day, -1, date.dte), n + 1
from dates
where n < 30
)
. . .

Answer 2

Use a tally table to generate the date and then query:
DECLARE #StartDate AS Date
SET #StartDate = '2018-10-30'
WITH N AS (
SELECT N
FROM (VALUES(NULL),(NULL),(NULL),(NULL),(NULL)) N(N)),
Tally AS(
SELECT TOP 30 ROW_NUMBER() OVER (ORDER BY (SELECT NULL)) -1 AS I
FROM N N1
CROSS JOIN N N2
CROSS JOIN N N3),
Dates AS(
SELECT DATEADD(DAY, I, #StartDate) AS [Date]
FROM Tally
)
SELECT u.member_id AS Member_id
,CAST(MAX(pe.executed_time) AS Date) AS Max_Date
,DATEDIFF(dd,#StartDate,MAX(pe.executed_time))*-1+1 AS R
,COUNT(*) AS F
,SUM(pi.Price) AS M
FROM [user] AS u
CROSS JOIN Dates D
LEFT JOIN purchase_entry AS pe
ON u.member_id=pe.member_id
LEFT JOIN purchase_item AS pi
ON pe.order_id=pi.order_id
WHERE pe.[status] = 'purchase'
AND pe.executed_time <= D.[Date]
ORDER BY Recency;