Find errors in a sequence - mysql

Question

I have an activity changelog of officers that become active/inactive.
OfficerID ChangeTo ChangeDate
1 active 2017-05-01
1 active 2017-05-02
1 inactive 2017-05-04
6 active 2013-09-09
6 inactive 2016-04-14
6 recruit 2016-06-22
6 active 2016-06-23
6 inactive 2017-04-30
In the case above, officer id 1 is active from 1st of May until the 4th of May.
This is essentially a 'housekeeping' task. The second row is not required and should be deleted. I would like to do this within a MySQL procedure that is linked an event on a schedule. I need a query that can identify these rows, but I'm not sure how.
In a previous system, I had looped through an ordered list and compared the current value against the previous row's value. I have read that loops in MySQL are not encouraged, so I'm trying to figure out how to do this with queries alone.
I tried the following:
SELECT
a.ActivityID, a.OfficerID, a.ChangeTo, a.ChangeDate
FROM
tbl_Officers_Activity as a
INNER JOIN tbl_Officers_Activity AS b
ON a.OfficerID = b.OfficerID
AND a.ChangeDate > b.ChangeDate
AND a.ChangeTo = b.ChangeTo
INNER JOIN tbl_Officers_Activity AS c
ON a.OfficerID = c.OfficerID
AND a.ChangeDate < c.ChangeDate
AND a.ChangeTo <> c.ChangeTo
ORDER BY
OfficerID,
ChangeDate;
I was hoping I could somehow embed the criteria I need into the joins, but I'm at a loss. any help would be greatly appreciated.

Answer 1

This is what you need
SQLFIddle Demo
select a1.officerid,a1.changedate,a1.changeto_a as changeto
From
(select a.officerid,a.changedate,max(a.changeto) as changeto_a,count(*) as rnk
from tbl_Officers_Activity a
inner join tbl_Officers_Activity b
on a.OfficerID=b.OfficerID
and a.ChangeDate>=b.ChangeDate
group by a.officerid,a.changedate) a1
left join
(select a.officerid,a.changedate,max(a.changeto) as changeto_b,count(*) +1 as rnk
from tbl_Officers_Activity a
inner join tbl_Officers_Activity b
on a.OfficerID=b.OfficerID
and a.ChangeDate>=b.ChangeDate
group by a.officerid,a.changedate) b1
on a1.officerid=b1.officerid
and a1.rnk=b1.rnk
where changeto_a = changeto_b
Explanation:
MySQL doesn't have row_Number function, so first I had to derive it. I used this query to get row_number, which is names as rnk in the query. Call the table a1.
(select a.officerid,a.changedate,max(a.changeto) as changeto_a,count(*) as rnk
from tbl_Officers_Activity a
inner join tbl_Officers_Activity b
on a.OfficerID=b.OfficerID
and a.ChangeDate>=b.ChangeDate
group by a.officerid,a.changedate)
Now as MySQL doesn't have LEAD function also, I derived it by using the above query again, and changing the rnk to rnk+1, calling it b1.
Now to replicate LEAD, I left joined a1 with b1
Now using a where clause to find same changeto, you can get your output.

Answer 2

My solution in the end was similar to what Utsav posted, however, after comparing his solution to mine, I found mine to be accurate, and his inaccurate.
For my solution, I create a temporary table for the initial ordered list with an auto incremental primary key. Then clone the temp table and join on itself with the primary pk equal to itself +1 and also equal to the status change.
my procedure ends up like this:
-- create temp table
DROP TABLE IF EXISTS tmp_act;
CREATE TEMPORARY TABLE IF NOT EXISTS tmp_act (
AutoID INT(11) NOT NULL AUTO_INCREMENT PRIMARY KEY,
ActivityID INT(11),
OfficerID INT(11),
ChangeTo text
);
-- temp data
insert into tmp_act (ActivityID,OfficerID,ChangeTo)
SELECT
a.ActivityID, a.OfficerID, a.ChangeTo
FROM
tbl_Officers_Activity as a
ORDER BY
OfficerID,
ChangeDate;
-- housekeeping
DROP TABLE IF EXISTS tmp_act1;
DROP TABLE IF EXISTS tmp_act2;
CREATE TEMPORARY TABLE IF NOT EXISTS tmp_act1 AS (SELECT * FROM tmp_act);
CREATE TEMPORARY TABLE IF NOT EXISTS tmp_act2 AS (SELECT * FROM tmp_act);
SELECT
a.AutoID, a.ActivityID, a.OfficerID, a.ChangeTo
FROM
tmp_act1 as a
INNER JOIN tmp_act2 AS b
ON a.OfficerID = b.OfficerID
AND a.ChangeTo = b.ChangeTo
AND a.AutoID = b.AutoID+1;