与 group by 结合使用时的 SQL 案例

Question

lets say I have a table with 2 columns: car_id, and selling_code

• if a car got sold at 2020, no matter how many times, it will have a single line in the table with the code 1.

• if a car got sold at 2019, no matter how many times, it will have a single line in the table with the code 2.

• if it got sold both at 2019 and 2020, it will have 2 lines, one with code 1 and one with code 2.

I want to create from this table a new table, in which there is only a single line for each id, and 2 columns: car_id and selling_text. if the car got sold only at 2020, the selling_text is '2020' else if the car got sold only at 2019, the selling_text is '2019' and if it got sold both at 2020 and 2019, the selling_text is 'both'

Answer 1

select car_id,
       case when count(distinct selling_code) = 2 then 'both'
            when max(selling_code) = 1 then '2020'
            else '2019'
       end as selling_text
from your_table
group by car_id

Answer 2

Hmmm . . . you seem to want:

select distinct car_id, year,
       (case when year = 2019 then 1 when year = 2020 then 2 end)
from t
where year in (2019, 2020);

This assumes that you actually have a year in the table. Although the question suggests that this is in the data, the question also explicitly describes the table with no such column.

If you want to be fancy, you can also use:

select distinct car_id, year, year - 2018 as code
from t
where year in (2019, 2020);