Use correlated subquery over multiple columns - mysql

Question

DB-Fiddle
CREATE TABLE logistics (
id int primary key,
campaign VARCHAR(255),
event_type VARCHAR (255),
date_offered VARCHAR (255),
date_ordered VARCHAR (255),
date_delivered VARCHAR (255),
date_recorded VARCHAR (255),
date_completed VARCHAR (255),
quantity_offered VARCHAR(255),
quantity_ordered VARCHAR(255),
quantity_delivered VARCHAR(255),
quantity_recorded VARCHAR(255),
quantity_completed VARCHAR(255)
);
INSERT INTO logistics
(id, campaign, event_type,
date_offered, date_ordered,
date_delivered, date_recorded, date_completed,
quantity_offered, quantity_ordered,
quantity_delivered, quantity_recorded, quantity_completed
)
VALUES
("1", "C001", "offered", "2019-04-10", NULL, NULL, NULL, NULL, "500", NULL, NULL, NULL, NULL),
("2", "C001", "ordered", NULL, "2019-04-16", NULL, NULL, NULL, NULL, "450", NULL, NULL, NULL),
("3", "C001", "stored", NULL, NULL, "2019-04-18", NULL, NULL, NULL, NULL, "465", NULL, NULL),
("4", "C001", "stored", NULL, NULL, NULL, "2019-04-20", NULL, NULL, NULL, NULL, "440", NULL),
("5", "C001", "stored", NULL, NULL, NULL, NULL, "2019-04-22", NULL, NULL, NULL, NULL, "445"),
("6", "C002", "offered", "2019-08-15", NULL, NULL, NULL, NULL, "600", NULL, NULL, NULL, NULL),
("7", "C002", "ordered", NULL, "2019-09-03", NULL, NULL, NULL, NULL, "700", NULL, NULL, NULL),
("8", "C002", "stored", NULL, NULL, "2019-09-05", NULL, NULL, NULL, NULL, "690", NULL, NULL),
("9", "C002", "stored", NULL, NULL, NULL, "2019-09-08", NULL, NULL, NULL, NULL, "692", NULL),
("10", "C003", "offered", "2019-10-24", NULL, NULL, NULL, NULL, "300", NULL, NULL, NULL, NULL),
("11", "C003", "ordered", NULL, "2019-10-28", NULL, NULL, NULL, NULL, "250", NULL, NULL, NULL),
("12", "C003", "stored", NULL, NULL, "2019-10-31", NULL, NULL, NULL, NULL, "320", NULL, NULL),
("13", "C004", "offered", "2019-11-05", NULL, NULL, NULL, NULL, "800", NULL, NULL, NULL, NULL),
("14", "C004", "ordered", NULL, "2019-11-14", NULL, NULL, NULL, NULL, "870", NULL, NULL, NULL),
("15", "C004", "stored", NULL, NULL, "2019-11-16", NULL, NULL, NULL, NULL, "740", NULL, NULL),
("16", "C005", "offered", "2019-12-17", NULL, NULL, NULL, NULL, "240", NULL, NULL, NULL, NULL),
("17", "C005", "ordered", NULL, "2020-01-12", NULL, NULL, NULL, NULL, "250", NULL, NULL, NULL),
("18", "C005", "stored", NULL, NULL, "2020-01-16", NULL, NULL, NULL, NULL, "226", NULL, NULL),
("19", "C006", "offered", "2020-01-09", NULL, NULL, NULL, NULL, "100", NULL, NULL, NULL, NULL),
("20", "C006", "ordered", NULL, "2020-01-23", NULL, NULL, NULL, NULL, "105", NULL, NULL, NULL),
("21", "C007", "offered", "2020-02-17", NULL, NULL, NULL, NULL, "900", NULL, NULL, NULL, NULL),
("22", "C008", "offered", "2020-02-25", NULL, NULL, NULL, NULL, "400", NULL, NULL, NULL, NULL);
The table above displays the purchasing process of different campaigns.
The first three process steps are defined by the event_types offered, ordered and stored.
Once a campaign has reached the process step stored the process continues with different dates called date_completed, date_recorded and date_delivered. Basically, those dates are sub_events from the event_type stored.
The hierarchy of those events is like this:
event_type = stored > ordered > offered
date = date_completed > date_recorded > date_delivered
Now, I want to extract the campaigns based on their highest event_type or - in case they already have reached the event_type stored - based on their highest date according to the defined hierarchy. The result should look like this:
campaign event_type date quantity
C001 stored 2019-04-22 445
C002 stored 2019-09-08 692
C003 stored 2019-10-31 320
C004 stored 2019-11-16 740
C005 stored 2020-01-16 226
C006 ordered 2020-01-23 105
C007 offered 2020-02-17 900
C008 offered 2020-02-25 400
With reference to this question I tried to modify the query like this:
SELECT
campaign,
event_type,
coalesce(date_offered, date_ordered) as main_event_date,
coalesce(date_delivered, date_recorded, date_completed) as sub_event_date,
coalesce(quantity_offered, quantity_ordered) as main_event_quantity,
coalesce(quantity_delivered, quantity_recorded, quantity_completed) as sub_event_quantity
FROM logistics lg
WHERE lg.id = (SELECT lg2.id
FROM logistics lg2
WHERE lg2.campaign = lg.campaign
ORDER BY field(lg2.event_type, 'stored', 'ordered', 'offered')
field(lg2.sub_event_date, 'date_completed', 'date_recorded', 'date_delivered')
LIMIT 1
);
However, I don't know how I should you use the field function once the campaign is in the event_type stored because there is no sub_event column for the dates?

Answer 1

You can use a subquery and group te data by campaign:
SELECT campaign,MAX(event_type) AS event_type,
CASE WHEN MAX(sub_event_date)>MAX(main_event_date) THEN MAX(sub_event_date) ELSE MAX(main_event_date) END AS event_date,
CASE WHEN MAX(sub_event_date)>MAX(main_event_date) THEN MAX(sub_event_quantity) ELSE MAX(main_event_quantity) END AS event_quantity
FROM (
SELECT
campaign,
event_type,
coalesce(date_offered, date_ordered) as main_event_date,
coalesce(date_delivered, date_recorded, date_completed) as sub_event_date,
coalesce(quantity_offered, quantity_ordered) as main_event_quantity,
coalesce(quantity_delivered, quantity_recorded, quantity_completed) as sub_event_quantity
FROM logistics lg) l
GROUP BY campaign

Answer 2

DB-Fiddle
The solution is as the following:
Step 1: Create an aggregated_sub_table called AGR_logistics which adds an additional column called sub_event_type to the original table:
CREATE TABLE AGR_logistics
SELECT
id,
campaign,
event_type,
(CASE
WHEN event_type = "stored" AND quantity_delivered IS NOT NULL THEN "delivered"
WHEN event_type = "stored" AND quantity_recorded IS NOT NULL THEN "recorded"
WHEN event_type = "stored" AND quantity_completed IS NOT NULL THEN "completed"
ELSE NULL END) AS sub_event_type,
date_offered,
date_ordered,
date_delivered,
date_recorded,
date_completed,
quantity_offered,
quantity_ordered,
quantity_delivered,
quantity_recorded,
quantity_completed
FROM logistics;
Step 2: Run a correleated subquery as described here over AGR_logistics:
SELECT
id,
campaign,
event_type,
sub_event_type,
coalesce(date_offered, date_ordered, date_delivered, date_recorded, date_completed) as event_date,
coalesce(quantity_offered, quantity_ordered, quantity_delivered, quantity_recorded, quantity_completed) as event_quantity
FROM AGR_logistics AGR_01
WHERE AGR_01.id = (SELECT AGR_02.id
FROM AGR_logistics AGR_02
WHERE AGR_02.campaign = AGR_01.campaign
ORDER BY
field(AGR_02.event_type, 'stored', 'ordered', 'offered'),
field(AGR_02.sub_event_type, 'completed', 'recorded', 'delivered')
LIMIT 1
);