Question

我有以下语句从Teradata DB中的表中提取日期，小时和用户数。。。

SELECT
    CAST(end_time AS DATE) AS end_date, 
    EXTRACT(HOUR FROM end_time) AS end_hour, 
    COUNT(users) AS total_users
FROM table
GROUP BY end_date, end_hour

使用extract（）函数时，我的结果集包含缺少的小时数，用户在24小时内没有活动...我想知道是否有任何技术可以解决我的结果集中缺少的小时数？

我无法创建要查阅的查找表，因为我没有必要的权限来在此数据库上创建表。

任何帮助将不胜感激！

Answer 1

sys_calendar.calendar生成请求的日期（根据需要更改范围）
WITH RECURSIVE生成小时数

with        recursive cte_hours (hr) 
            as
            (
                            select  0      from (select 1) t(c) 
                union all   select  hr + 1 from cte_hours where hr < 23
            )

select      c.calendar_date             as dt
           ,h.hr                        as hr
           ,zeroifnull(t.total_users)   as total_users

from                    sys_calendar.calendar   as c 

            cross join  cte_hours               as h

            left join  (select      cast(end_time as date)      as end_date
                                   ,extract(hour from end_time) as end_hour
                                   ,count(users)                as total_users

                        from        mytable t 

                        group by    end_date
                                   ,end_hour
                        ) t

            on          t.end_date  = c.calendar_date 
                    and t.end_hour  = h.hr 

where       c.calendar_date between current_date - 10 and current_date     

order by    dt,hr
;

对于@GordonLinoff

select 0

0

select 1

1

select 0
union all
select 1

[3888] UNION，INTERSECT或MINUS的SELECT必须引用一个表格。

select 0 from (select 1 as c) t
union all
select 1 from (select 1 as c) t

0
1

或

select 0 from (select 1) t(c)
union all
select 1 from (select 1) t(c)

0
1

Answer 2

如果您想要数据库中所有日期的所有小时数，那么您可以使用cross join生成行，然后使用left join来生成结果：

SELECT d.end_date, 
       EXTRACT(HOUR FROM end_time) AS end_hour, 
        COUNT(t.users) AS total_users
FROM (select distinct CAST(end_time AS DATE) AS end_date from table) d CROSS JOIN
     (select distinct EXTRACT(HOUR FROM end_time) AS end_hour from table) h LEFT JOIN
     table t
     ON t.end_date = d.end_date and t.end_hour = d.end_hour
GROUP BY e.end_date, h.end_hour;

如果未表示所有小时数，您可以使用明确的列表：

SELECT d.end_date, 
       EXTRACT(HOUR FROM end_time) AS end_hour, 
        COUNT(t.users) AS total_users
FROM (select distinct CAST(end_time AS DATE) AS end_date from table) d CROSS JOIN
     (select * from (select 0 as end_hour) t UNION ALL 
      select * from (select 1 as end_hour) t UNION ALL
      . . .
     ) h LEFT JOIN
     table t
     ON t.end_date = d.end_date and t.end_hour = d.end_hour
GROUP BY e.end_date, h.end_hour;

Teradata - 如何使用extract（）函数计算时间戳中缺少的小时数

2 个答案: