mySQL largest value by unique entry - mysql

Question

I have tried to solve the following problem for the last couple of hours and could not find anything that pointed me in the right direction on Google or Stackoverflow. I believe that this could be a similar problem, but I did not really understand what the author wanted to achieve, hence I am trying it with my own concrete example:
I have a table that basically tracks prices of different products over time:
+------------+--------+----------+
| Product_id | Price | Time |
+------------+--------+----------+
| 1 | 1.30 | 13:00:00 |
| 1 | 1.10 | 13:30:00 |
| 1 | 1.50 | 14:00:00 |
| 1 | 1.60 | 14:30:00 |
| 2 | 2.10 | 13:00:00 |
| 2 | 2.50 | 13:30:00 |
| 2 | 1.90 | 14:00:00 |
| 2 | 2.00 | 14:30:00 |
| 3 | 1.45 | 13:00:00 |
| 3 | 1.15 | 13:30:00 |
| 3 | 1.50 | 14:00:00 |
| 3 | 1.55 | 14:30:00 |
+------------+--------+----------+
I would now like to query the table so that the rows with max. Price for each product are returned:
+------------+--------+----------+
| Product_id | Price | Time |
+------------+--------+----------+
| 1 | 1.60 | 14:30:00 |
| 2 | 2.50 | 13:30:00 |
| 3 | 1.55 | 14:30:00 |
+------------+--------+----------+
Also, in case of duplicates, i.e. if there is a max. Price at two different points in time, it should only return one row, preferably the one with the smallest value of time.
I have tried MAX() and GREATEST(), but could not achieve the desired outcome to show the wanted values for each product. Efficiency of the query is not the most important factor, but I have about 500 different products with several million rows of data, hence splitting the table by unique product did not seem like an appropriate solution.

Answer 1

Group the data product id and pick the max price and max time
select t1.product_id,t1.price,min(t1.time) as time from your_table t1
inner join (
select Product_id,max(price)as price from
your_table group by Product_id
) t2 on t1.Product_id=t2.Product_id and t1.price=t2.price group by t1.product_id
Sql Fiddle Example:
http://sqlfiddle.com/#!9/020c3/9

Answer 2

http://sqlfiddle.com/#!9/020c3/1
SELECT p.*
FROM prices p
LEFT JOIN prices p1
ON p.product_id = p1.product_id
AND p.time<p1.time
WHERE p1.product_id IS NULL
If you need maximum price to get you can:
http://sqlfiddle.com/#!9/020c3/6
SELECT p.*
FROM prices p
LEFt JOIN prices p1
ON p.product_id = p1.product_id
AND p.price<p1.price
WHERE p1.product_id IS NULL;
And the last approach since I didn't get the goal from the beggining:
http://sqlfiddle.com/#!9/ace04/2
SELECT p.*
FROM prices p
LEFt JOIN prices p1
ON p.product_id = p1.product_id
AND (
p.price<p1.price
OR (p.price=p1.price AND p.time<p1.time)
)
WHERE p1.product_id IS NULL;

Answer 3

This solution is assuming the existence of an additional my_table.id column that needs to be used in case there are duplicate values for (Product_id, price, time) in your table. id is assumed to be a unique value in the table.
SELECT *
FROM my_table t1
WHERE NOT EXISTS (
SELECT *
FROM my_table t2
WHERE t1.Product_id = t2.Product_id
AND ((t1.price < t2.price) OR
(t1.price = t2.price AND t1.time > t2.time) OR
(t1.price = t2.price AND t1.time = t2.time AND t1.id > t2.id))
)
Alternatively, the predicate on price and time could also be expressed using a row value expression predicate (not sure if it's more readable, as t1 and t2 columns are mixed in a each row value expression):
SELECT *
FROM my_table t1
WHERE NOT EXISTS (
SELECT *
FROM my_table t2
WHERE t1.Product_id = t2.Product_id
AND (t1.price, t2.time, t2.id) < (t2.price, t1.time, t1.id)
)