Merging 2 columns from 2 different tables into a single one

Question

I have 2 tables:

• 1 for the visit of persons in a shop
• 1 for purchase made by these visitors

Therefore a buyer has to be a visitor as well.

Here are the data that I have:

PURCHASE_TABLE

day         user_id_purchase  item_id_purchase  Type    
----------  ----------------  ----------------  --------    
26/05/2016  AAA               47332             Purchase        
19/05/2016  BBB               46523             Purchase        

VISIT_TABLE

day         user_id_visit  Type
----------  -------------  -----
18/06/2016  AAA            Visit    
26/05/2016  AAA            Visit    
19/05/2016  BBB            Visit    
18/05/2016  CCC            Visit

Here is what I would like to have: day user_id type item_id from both tables

With the results:

day         user_id  type      item_id
----------  -------  --------  -------
18/06/2016  AAA      Visit  
26/05/2016  AAA      Visit  
19/05/2016  BBB      Visit  
18/05/2016  CCC      Visit  
26/05/2016  AAA      Purchase  47332
19/05/2016  BBB      Purchase  46523

Nevertheless I am unable to do it. The result that I got is a multiplication of the number of rows so far: I got 4 rows(visit)*2 rows (purchase) instead of 4 rows+2 rows. Indeed, I got each purchase and each click with the purchase…

Here is the query that I use:

SELECT
  visits.user_id,
  coalesce(visits.day_visit, purchases.day_purchase) AS day,
  coalesce(visits.type, purchases.type) AS type,
  purchases.item_id_purchase
FROM
  (SELECT DISTINCT
     day          AS day_visit,
     user_id AS user_id,
     'Visit'    AS type
   FROM visit 
    WHERE DAY >= '2016-01-01') visits,
  (SELECT DISTINCT
     day          AS day_purchase,
     user_id_slow AS user_id,
     item_id      AS item_id_purchase,
     'Purchase'      AS type
   FROM purchase
   WHERE AND Day >= '2016-05-02') purchases
WHERE visits.user_id_display = purchases.user_id

more ore less similar to what I found here: How can I merge the columns from two tables into one output? I think that it doesn't work since I am using columns which are different between the 2 tables.

I tried with JOINs (both INNER AND LEFT) as well and it did not bring better results.

Do you have any idea on how I can get the result I wish?

Thanks,

JP

Answer 1

You want all the rows of both tables displayed, which you can do with UNION and faking the missing column from the Visit table.

SELECT day_date, user_id, type, NULL AS item_id
FROM Visit
UNION 
SELECT day_date, user_id, type, item_id
FROM Purchase
ORDER BY  type DESC, user_id ASC , day_date ASC

Which gives the result

day_date     user_id type        item_id
------------ ------- ----------  ------- 
"18-06-2016" "AAA"   "Visit"     "NULL"
"26-05-2016" "AAA"   "Visit"     "NULL"
"19-05-2016" "BBB"   "Visit"     "NULL"
"18-05-2016" "CCC"   "Visit"     "NULL"
"26-05-2016" "AAA"   "Purchase"  "47332" 
"19-05-2016" "BBB"   "Purchase"  "46523"