Using DATEDIFF based on the max date of each instance of another column

Question

I have a status table the contains an orderNO, Insert_Date, and status. My goal is to determine the time duration between status changes. The issue is that if whoever is entering the order clicks the status change buttons multiple times, I'll have multiple instances of each status. I pulled an example of a particular orderNO below:

+------------+---------+--------------------------+
|  orderNO   | status  |       insert_date        |
+------------+---------+--------------------------+
| OBJV107522 | ENTERED | 2/4/2019 11:44:45.800 AM |
| OBJV107522 | ENTERED | 2/4/2019 11:45:14.410 AM |
| OBJV107522 | ENTERED | 2/4/2019 11:45:14.597 AM |
| OBJV107522 | ENTERED | 2/4/2019 11:45:14.833 AM |
| OBJV107522 | OPEN    | 2/4/2019 12:27:15.710 PM |
| OBJV107522 | ENTERED | 2/4/2019 12:36:39.327 PM |
| OBJV107522 | ENTERED | 2/4/2019 12:36:39.920 PM |
| OBJV107522 | OPEN    | 2/4/2019 2:34:58.957 PM  |
| OBJV107522 | ENTERED | 2/4/2019 2:35:07.817 PM  |
| OBJV107522 | OPEN    | 2/4/2019 3:50:04.393 PM  |
+------------+---------+--------------------------+

I would ideally like to see the minutes elapsed between a status change. The output would need to look like this.

• The first row should be the max instance of all of the "ENTERED"s before the status changes minus the max of the first instances of "OPEN" before the status changes.
• The next row should be the max of the first "OPEN" instances (before the status changes back to "ENTERED" minus the max instance of the second "ENTERED".

    +------------+------------+-----------------+
    |  orderNO   | New_status | minutes_elapsed |
    +------------+------------+-----------------+
    | OBJV107522 | OPEN       |              42 | 4th row - 5th row of the original data
    | OBJV107522 | ENTERED    |               9 | 5th row-7th row of the original data
    | OBJV107522 | OPEN       |             118 |
    | OBJV107522 | ENTERED    |               0 |
    | OBJV107522 | OPEN       |              75 |
    +------------+------------+-----------------+

I'm getting no where trying this on my own. I'm starting to convince myself it's not possible. Please let me know of any suggestions.

Answer 1

Maybe something like this to take advantage of windowing functions.

WITH CTE AS(
    SELECT *, 
        LAG( status) OVER( PARTITION BY orderNO ORDER BY insert_date) AS Previous_Status, 
        LAG( insert_date) OVER( PARTITION BY orderNO ORDER BY insert_date) AS Previous_Date
    FROM OrderStatus
)
SELECT orderNO,
    status,
    DATEDIFF(mi, Previous_Date, insert_date) AS minutes_elapsed,
    ROUND(DATEDIFF(ss, Previous_Date, insert_date)/60., 0) AS minutes_elapsed2 /*This actually matches your expected results*/
FROM CTE
WHERE status <> Previous_Status;

Answer 2

DECLARE @t TABLE (orderNO VARCHAR(20),status  VARCHAR(20), insert_date DATETIME)

INSERT INTO @t VALUES
(' OBJV107522 ',' ENTERED ',' 2/4/2019 11:44:45.800 AM '),
(' OBJV107522 ',' ENTERED ',' 2/4/2019 11:45:14.410 AM '),
(' OBJV107522 ',' ENTERED ',' 2/4/2019 11:45:14.597 AM '),
(' OBJV107522 ',' ENTERED ',' 2/4/2019 11:45:14.833 AM '),
(' OBJV107522 ',' OPEN    ',' 2/4/2019 12:27:15.710 PM '),
(' OBJV107522 ',' ENTERED ',' 2/4/2019 12:36:39.327 PM '),
(' OBJV107522 ',' ENTERED ',' 2/4/2019 12:36:39.920 PM '),
(' OBJV107522 ',' OPEN    ',' 2/4/2019 2:34:58.957 PM  '),
(' OBJV107522 ',' ENTERED ',' 2/4/2019 2:35:07.817 PM  '),
(' OBJV107522 ',' OPEN    ',' 2/4/2019 3:50:04.393 PM  ');

SELECT t2.orderNO, t2.status, t2.insert_date
    , minutes_elapsed  = MAX(DATEDIFF(SECOND, t.insert_date, t2.insert_date)/60)
FROM @t as t
CROSS APPLY (
    SELECT insert_date = MIN(t1.insert_date) 
    FROM @t as t1
    WHERE t1.status != t.status 
        and t1.insert_date > t.insert_date
) as tm
INNER JOIN @t as t2 ON t2.insert_date = tm.insert_date 
GROUP BY t2.orderNO, t2.status, t2.insert_date
ORDER BY t2.insert_date