Re: Complicated GROUP BY

>
> Dear All,
>
> I have the following problem with grouping: I want to know the maximum
> in a group as well as the maximal element. Example:
>
> I have a table products_providers:
> product | provider | criteria_1 | criteria_2
>
> I have a number of products, each of them from a several providers.
> Each product is described by two numeric values. I can easily select
> the best value for each product by a given criteria, like:
>
> select product, max(criteria_1) from products_providers group by
> product;
>
> but I need to know the best-scoring provider as well.
>
> Result I need should look like:
> product | best_provider_1 | best_criteria_1 | best_provider_2 |
> best_criteria_2
>
> If it counts results may be split into two tables: one for the first
> and the other for the second criteria
>
> Can you help me with a painless solution?

Is something like this what you're after?

select * from products_proivders
order by criteria_1 desc limit 1

You can get the best providers for both criteria using union like this:

select * from (
select 'best_criteria_1' as name, product, provider, criteria_1, criteria_2
from products_proivders order by criteria_1 desc limit 1 ) x
union
select * from (
select 'best_criteria_2' as name, product, provider, criteria_1, criteria_2
from products_proivders order by criteria_2 desc limit 1 ) y