Question

I have a table like this:

| X | Y | Data |
----------------
| 1 | 2 | Hi   |
| 3 | 4 | Hey  |
| 1 | 2 | New! |
| 1 | 2 | New! |
| 1 | 2 | New! |

A little back story:

I am trying to create a table (HTML) using the co-ordinates specified in my table with the data in the cell body. Because the data isn’t specifically gotten by myself and comes from the users themselves (Grease monkey script pulling data from the site) I need to group the data by X and Y then group by the Data column and pick out the most occurring one.This is so I get legitimate data as they can spoof the data without me able to do anything about it.

The trouble being is finding a select query that will produce some sort of array like the following from the above table.

Array ( 
        [0] = Array ( 
           'x' => 1,
           'y' => 2,
           'Data' => "New!"   
         )
        [1] = Array ( 
           'x' => 3,
           'y' => 4,
           'Data' => "Hey"   
         )
      )

I did try a few variations but none worked so I’m checking here to see if anyone could shed some light on my problem.

Thanks!

Answer 1

Does this work for you?

SELECT T3.X, T3.Y, MAX(T3.Data) AS Data
FROM
(
    SELECT X, Y, MAX(cnt) AS max_cnt
    FROM
    (
        SELECT X, Y, Data, COUNT(*) AS cnt
        FROM yourtable
        GROUP BY X, Y, Data
    ) T1
    GROUP BY X, Y
) T2
JOIN
(
    SELECT X, Y, Data, COUNT(*) AS cnt
    FROM yourtable
    GROUP BY X, Y, Data
) T3
ON T2.X = T3.X AND T2.Y = T3.Y AND T2.max_cnt = T3.cnt
GROUP BY T3.X, T3.Y

You may also want to consider creating a view:

CREATE VIEW yourtable_counts AS
SELECT X, Y, Data, COUNT(*) AS cnt
FROM yourtable
GROUP BY X, Y, Data

Then you can simplify the query:

SELECT T3.X, T3.Y, MAX(T3.Data) AS Data
FROM
(
    SELECT X, Y, MAX(cnt) AS max_cnt
    FROM yourtable_counts T1
    GROUP BY X, Y
) T2
JOIN yourtable_counts T3
ON T2.X = T3.X AND T2.Y = T3.Y AND T2.max_cnt = T3.cnt
GROUP BY T3.X, T3.Y