Question

I have a table that stores car prices with a structure like:

+-----------------+--------------+------+-----+---------+----------------+
| Field           | Type         | Null | Key | Default | Extra          |
+-----------------+--------------+------+-----+---------+----------------+
| id              | int(11)      | NO   | PRI | NULL    | auto_increment |
| model           | varchar(255) | YES  |     | NULL    |                |
| make            | varchar(255) | YES  |     | NULL    |                |
| year            | varchar(255) | YES  |     | NULL    |                |
| avg_price       | decimal(8,2) | YES  |     | NULL    |                |
| median_price    | decimal(8,2) | YES  |     | NULL    |                |
| created_at      | datetime     | YES  |     | NULL    |                |
| updated_at      | datetime     | YES  |     | NULL    |                |
+-----------------+--------------+------+-----+---------+----------------+

The data for a given year might be inserted at different times.

For example model 'Honda', make 'Accord' might have results like:
+------+-----------+--------------+--------+---------------------+
| year | avg_price | median_price | model  | created_at          |
+------+-----------+--------------+--------+---------------------+
| 1992 |   2431.29 |      2000.00 | accord | 2012-02-23 17:31:41 |
| 1993 |   2609.13 |      2195.00 | accord | 2012-02-23 17:31:44 |
| 1994 |   2858.81 |      2400.00 | accord | 2012-02-23 17:31:44 |
| 2000 |   4771.99 |      4450.00 | accord | 2012-02-23 17:31:46 |
| 2001 |   5260.16 |      5000.00 | accord | 2012-02-23 17:31:46 |
| 2000 |   4860.19 |      4795.00 | accord | 2012-08-15 06:09:52 |
| 2001 |   5071.49 |      4990.00 | accord | 2012-08-15 06:09:52 |
| 2002 |   5872.80 |      5795.00 | accord | 2012-08-15 06:09:52 |
| 2003 |   7521.44 |      7950.00 | accord | 2012-08-15 06:09:52 |
| 2004 |   8348.19 |      8495.00 | accord | 2012-08-15 06:09:52 |

I would like to retrieve all honda accord data that is the latest for that year of car make.

so in the above example, I would like to retrieve data from 2012-08-15 06:09:52 for years 2000,2001,2002,2003,2004
but the older years would be from the date 2012-02-23 17:31:41

select year,avg_price,median_price,model,created_at 
from car_prices 
where make='honda' and    model= 'accord' group by year asc

The above query gets distinct data for each year but not the last record inserted for each year.

Any ideas how to get distinct data for each year as well as the latest?

Answer 1

For both performance and guarnanteed behaviour, you should create a lookup and join on that.

In your case you want to find the most recent created_at value for any given make, model, year group. The sub-query in the code below does that.

Then you join that back on your original data again, finding only the records that have those make, model, year, created_at values.

This does mean that it you have more than one record with the same make, model, year, created_at values, you will get multiple results for that make, model, year.

Ensure that you have an index covering (make, model, year, created_at) to make the search for the most recent created_at quick, as well as the join.

SELECT
  *
FROM
  car_prices
INNER JOIN
(
  SELECT
    make, model, year, MAX(created_at) as created_at
  FROM
    car_prices
  GROUP BY
    make, model, year
)
  AS most_recent
    ON  car_prices.make       = most_recent.make
    AND car_prices.model      = most_recent.model
    AND car_prices.year       = most_recent.year
    AND car_prices.created_at = most_recent.created_at
WHERE
      most_recent.make  = 'honda'
  AND most_recent.model = 'accord'