有条件地找到明显的计数

Question

使用下面的表table1跟踪租车站的汽车移动

+---------+---------+----------+--------+------+-------+-------+--------+
| station |   make  |   model  |  start |  end |  kms  |  time |  regno |
+---------+---------+----------+--------+------+-------+-------+--------+
|    1111 |  toyota |  camry   |  ca    |  mh  |  1200 |    12 |   2222 |
|    1111 |  toyota |  camry   |  ca    |  mg  |  1300 |    14 |   2233 |
|    1111 |  honda  |  accord  |  ab    |  mx  |  1400 |    12 |   2255 |
|    1111 |  honda  |  accord  |  ab    |  mx  |  1400 |    12 |   2255 |
|    1122 |  toyota |  corolla |  ab    |  mg  |   800 |     8 |   2244 |
|    1133 |  honda  |  accord  |  ab    |  mx  |   900 |     7 |   2255 |
|    1133 |  honda  |  accord  |  ab    |  mx  |   900 |     7 |   2277 |
+---------+---------+----------+--------+------+-------+-------+--------+

需要为每个电台查找不同的汽车（regno），与honda和non-honda不同的汽车，如下面的预期输出

+---------+----------------------+--------------------------+------------------------------+
| station |  distinct_cars_count |  make_honda_distinct_cnt |  make_non_honda_distinct_cnt |
+---------+----------------------+--------------------------+------------------------------+
|    1111 |                    3 |                        1 |                            2 |
|    1122 |                    1 |                        0 |                            1 |
|    1133 |                    2 |                        2 |                            0 |
+---------+----------------------+--------------------------+------------------------------+

我可以在下面找到不同的汽车

select
 count(distinct regno) as distinct_cars_count 
from table1
group by station

我需要帮助找到make_honda_distinct_cnt这是与make = 'honda'和make_non_honda_distinct_cnt不同的汽车，这些汽车将是make <> 'honda'

的不同汽车

Answer 1

您可以使用filter()子句使用条件聚合：

select count(distinct regno) as distinct_cars_count , 
       count(distinct regno) filter (where make = 'honda') as make_honda_distinct_cnt, 
       count(distinct regno) filter (where make <> 'honda') as make_non_honda_distinct_cnt
from table1
group by station

Answer 2

您可以在计数中使用case语句：

select  station,
        count(distinct regno) as distinct_cars_count,
        count(distinct case when make = 'honda' then regno end) as make_honda_distinct_cnt,
        count(distinct case when make <> 'honda' then regno end) as make_non_honda_distinct_cnt,
from    table1
group by station