Question

I would like to generate a table (or query result) in this form

+---------------------+---------------------+
| Email               | Favourite Brand ID  |
+---------------------+---------------------+
| [e-mail]  |                  89 |
+-                   -+-                   -+
| [e-mail]   |                 193 |
+-                   -+-                   -+

I have managed to write a query that generates a list of unique brand ID’s with customer email addresses and the number of times that customer has purchased that brand. The results look something like this:

+---------------------+-----------+---------------+
| Email               | Brand ID  | CountOfOrders |
+---------------------+-----------+---------------+
| [e-mail]  |        89 |            10 |
+-                   -+-         -+-             -+
| [e-mail]   |       193 |            32 |
+-                   -+-         -+-             -+
| [e-mail] |        20 |             2 |
+-                   -+-         -+-             -+
| [e-mail] |        47 |             5 |
+-                   -+-         -+-             -+

Obviously [e-mail] has purchased from BrandID 20 twice and BrandID 47 5 times which is why they appear twice. Most customers have purchased from more than one brand.

From this information how can I construct a query to get the brand ID they have purchased from the most? I have tried the following but it just times out:

SELECT [table1].Email, [table1].Brand, [table1].CountOfBrand
FROM [Customer Brand Purchases] AS [table1]
GROUP BY [table1].Email, [table1].Brand, [table1].CountOfBrand
WHERE [table1].CountOfBrand=(
    SELECT TOP 1 [table2].CountOfBrand 
    FROM [Customer Brand Purchases] AS [table2] 
    WHERE [table2].Email = [table1].Email 
    ORDER BY [table2].CountOfBrand DESC
);

Oh and I have to use Microsoft Access, unfortunately. Thanks.

Answer 1

I am reluctant to answer this as the thought of even assisting in the development of a database with a table name [Customer Brand Purchases] makes me feel a little bit sick.

My Access SQL is a little rusty but I am 98% certain this will work:

SELECT   CBP.Email, CBP.Brand AS [Favourite Brand ID]
FROM    [Customer Brand Purchases] AS CBP
        INNER JOIN
        (   SELECT  [Email], MAX(CountofBrand) AS [MaxCountofBrand]
            FROM    [Customer Brand Purchases]
            GROUP BY [Email]
        ) AS [MaxCBP]
            ON CBP.Email = MaxCBP.Email
            AND CBP.CountOfBrand = MaxCBP.MaxCountOfBrand

The only draw back is that if a particular customer has ordered 2 brands the same amount of times then it will return 2 rows. You would need additional subqueries with MAX statements in to resolve this.

EDIT/ADDENDUM:

If it is ABSOLUTELY imperative the query returns 1 result per email address then you need to allow for the scenario where a particular email address has purchased 2 brands and equal amount of times there is no way to establish which or these is favourite as they are joint favourite. If it were me I would deal with this at application level, and Concatenate favourite brands into one string. However, it can be done in SQL just be aware that one or more brands could be hidden using this:

SELECT   CBP.Email, CBP.Brand AS [Favourite Brand ID]
FROM    [Customer Brand Purchases] AS CBP
        INNER JOIN
        (   SELECT  CBP.Email, MAX(CBP.Brand) AS MaxBrand
            FROM    [Customer Brand Purchases] AS CBP
                    INNER JOIN
                    (   SELECT  [Email], MAX(CountofBrand) AS MaxCountofBrand
                        FROM    [Customer Brand Purchases]
                        GROUP BY [Email]
                    ) AS MaxCBP
                        ON CBP.Email = MaxCBP.Email
                        AND CBP.CountOfBrand = MaxCBP.MaxCountOfBrand
            GROUP BY Email
        ) AS MaxCBP
            ON CBP.Email = MaxCBP.Email
            AND CBP.Brand = MaxCBP.Brand