我有一个数据库,其中有5个日期时间,它们是文本数据类型。值是4/15/12 1:47
(FT型),9/8/12 20:02
(FT型),5/10/13 22:21
(FT型),3/3/13 4:46
(FT型)和2/9/12 4:19
( NL型)。我想计算每小时发生的事件数,然后将其除以总事件数。
由于该列是文本数据类型,所以我用以下代码选择小时:hour(str_to_Date(order_placer_placed_time,'%m/%d/%Y %H:%i')
。
我尝试过的代码是
select (
(select count(*) from col where type = 'ft' and hour(str_to_Date(time,'%m/%d/%Y %H:%i')) = 4
group by type)
/count(*)) as 'FT %',
hour(str_to_Date(time,'%m/%d/%Y %H:%i')) as hour
from col
group by hour
order by hour;
我需要将子查询中的4
更改为每天的所有小时段,以便返回一个结果,该结果显示出'ft'个人的百分比除以总个人的百分比。
我不确定如何使子查询中的4
动态返回所要查找的结果。
最终输出应如下所示:
1 100%
4 50%
20 100%
22 100%
答案 0 :(得分:2)
也许您只想要这样的东西
drop table if exists col;
create table col
(type varchar(3),ts varchar(20));
insert into col values
('ft','1/2/2019 1:47'),('ft','1/2/2019 20:02'),('ft','1/2/2019 22:21'),('ft','1/2/2019 1:47'),('ft','1/2/2019 4:19'),
('ft','1/3/2019 1:47'),('nl','1/3/2019 4:19');
select hour(str_to_Date(ts,'%m/%d/%Y %H:%i')) as hour,
sum(case when type = 'ft' then 1 else 0 end) obsft,
count(*) obsall,
sum(case when type = 'ft' then 1 else 0 end) /
count(*) * 100 as perft
from col
group by hour(str_to_Date(ts,'%m/%d/%Y %H:%i'))
union
select 25,
sum(case when type = 'ft' then 1 else 0 end) obsft,
count(*) obsall,
sum(case when type = 'ft' then 1 else 0 end) /
count(*) * 100 as perft
from col;
+------+-------+--------+----------+
| hour | obsft | obsall | perft |
+------+-------+--------+----------+
| 1 | 3 | 3 | 100.0000 |
| 4 | 1 | 2 | 50.0000 |
| 20 | 1 | 1 | 100.0000 |
| 22 | 1 | 1 | 100.0000 |
| 25 | 6 | 7 | 85.7143 |
+------+-------+--------+----------+
5 rows in set (0.00 sec)