How to deal with blanks in database raw data?

Question

I've got raw transaction data that has a Transaction Number and a Card Number. Unfortunately, the card number is not repeated on every row of the transaction, so there are blanks. I want to summarise the data by Card Number, so I need to fill the blanks (I think).

The way I'm dealing with this problem, is to create a unique lookup of Transaction Number and Card Number, match the Card Number back onto the raw data, then summarize by Card Number.

I've attached a little diagram:

Any help to improve this process would be appreciated. Getting rid of the lookup step would be nice!

Answer 1

Just an example but should work:

  SELECT r.cardId AS `Card Number`, 
         COUNT(*) AS `# of Transactions`, 
         SUM(t.sales) AS `Total Sales`
    FROM transactions t
    JOIN (SELECT transactionId, cardId 
            FROM transactions
           WHERE cardId IS NOT NULL OR 
                 cardId != ''
         ) AS r
      ON r.transactionId = t.transactionId
GROUP BY t.transactionId

Basically what I am doing is:

• On the outer select we grab all the results grouped by the transactionId this way we can easily count the number of transaction and sum the total of sales.
• Then on the join I am basically filtering so we grab the cardId from a transaction that does not have it null or empty.

And the result should be what you described.

CARD NUMBER     # OF TRANSACTIONS   TOTAL SALES
       5845                     3           141
       3957                     5           177

Live DEMO

---

For the raw+matching result:

SELECT t.transactionId,
       r.cardId, 
       t.sales
  FROM transactions t
  JOIN (SELECT transactionId, cardId 
          FROM transactions
         WHERE cardId IS NOT NULL OR 
               cardId != ''
       ) AS r
    ON r.transactionId = t.transactionId

Live DEMO

---

To update the empty results you could use an UPDATE like this:

UPDATE transactions t
  JOIN (SELECT transactionId, cardId 
          FROM transactions
         WHERE cardId IS NOT NULL OR 
               cardId != ''
       ) AS m
    ON m.transactionId = t.transactionId
   SET t.cardId = m.cardId 
 WHERE t.cardId IS NULL OR 
       t.cardId = ''

Live DEMO