Question

I have two tables, a Countries table and a Weather table. I would like to retrieve all of the names of countries where it has not rained within the last 15 days.

The weather table has a column called “DayNum”, which goes from 1 -> infinity and increases by 1 on each day, it is unique. This table also has a column called “Rain” which is just a bit boolean value of 0 or 1.

Also, not all Countries were added on the same day, so the max DayNum will be different for each country.

Examples of tables below (data is snipped for readability):

Countries:

    ID     Name
     1      USA
     2      Cananda
     3      Brazil

Weather

    ID    Country_id    DayNum    Rain
     1        1           1         0
     2        1           2         0
     3        1           3         1

Here is my current attempt at a query (been working on this for days):

    SELECT countries.name, weather.daynum
    FROM countries INNER JOIN weather ON countries.id = weather.country_id
    GROUP BY countries.name
    HAVING weather.daynum > (MAX(weather.day_num) - 15) AND SUM(weather.rain) = 0;

I think this should work, but I’m having serious performance issues. The actual query I need to write deals with different data (same exact concept) and millions of rows. This query seems to get slower at an exponential rate.

Can anyone offer any advice?

Another idea I had was to somehow limit the JOIN to only grab the top 15 records (whilst ORDERing BY weather.day_num), but I Haven’t found a way to do this within a JOIN (if it’s even possible).

Answer 1

You’re not interested in the amount of rain, just whether it exists, so…

select * from countries
left join
(
        select weather.country_id 
        from weather 
            inner join 
            (select country_id, MAX(daynum) as maxdaynum from weather group by country_id) maxday
                on weather.country_id = maxday.country_id
                and weather.daynum>maxday.maxdaynum-3
                where rain=1
        ) rainy
on countries.id = rainy.country_id
where country_id is null    

I presume you’ve already indexed your tables appropriately