Question

Supose I have a table with several columns.
I know how to find duplicates, and then to keep just one record with this expression, adding a combinated index…

ALTER IGNORE TABLE part3 ADD UNIQUE INDEX (gtu, region, trait, pop, author, risk)

…but I would like to be able to control which will be the deleted row; not just the one with the “first row of rows with duplicates on a unique key“, as stated here in the mysql doc.
I mean, i exactly want to specify that the non-deleted row must be the one with the minor value in column field called zvalue.

Any suggestion?? Is this possible?

Thanks!

Answer 1

Instead of trying to delete what you don’t need, try to think of it as a problem whereby you want to pick the ones you want and ignore the rest; like so:

CREATE TABLE `other_table` LIKE `part3`;
INSERT INTO `other_table`
SELECT `part3`.*
FROM (SELECT gtu, region, trait, pop, author, risk, MAX(zvalue) AS max_zval
    FROM `part3`
    GROUP BY gtu, region, trait, pop, author, risk
) `tmp`
INNER JOIN `part3` USING (gtu, region, trait, pop, author, risk)
WHERE `part3`.zvalue = `tmp`.max_zval;

That should create a table other_table that satisfies the unique constraint in your data; only the rows with the highest zvalue are kept, the others removed (skipped).