Question

我有一个程序可以定期将用户移动数据连接的详细信息（信号强度，网络技术类型等）记录到PostgreSQL中的单个表中。表中的相关字段是;

networkmedium.id
networkmedium.date_time
networkmedium.medium_type

示例表可能看起来像;

0  |  2013-07-26 08:00:01  | 3G
1  |  2013-07-26 08:00:02  | GPRS
2  |  2013-07-26 08:00:06  | 3G
3  |  2013-07-26 08:00:10  | 3G
4  |  2013-07-26 08:00:16  | GPRS

我试图从这些数据中得出在特定媒体类型（即GPRS，3G等）上花费的总持续时间，并且我遇到了困难。

在上面的例子中，我需要r1和r0，r2和r1，r3和r1，r4和r3之间的时差，然后将它们加在一起按medium_type分组，但我陷入了最好的状态关于这个。

提前致谢，崔佛

Answer 1

试试这个：

with CTE1 as
(
    select
        *,
        lead(date_time) over (order by date_time asc) as next_date_time
    from networkmedium
    order by date_time asc
)
select C1.medium_type, sum(date_part('second', C1.next_date_time - C1.date_time))
from CTE1 as C1
group by C1.medium_type

MEDIUM_TYPE |    SUM
--------------------
GPRS        |      4
3G          |     11

SQL FIDDLE EXAMPLE

更复杂的方法

with CTE1 as
(
    select
        *,
        lag(medium_type) over (order by date_time asc) as prev_medium_type
    from networkmedium
    order by date_time asc
), CTE2 as
(
    select *, row_number() over(order by date_time) as row_num
    from CTE1
    where prev_medium_type <> medium_type or prev_medium_type is null
)
select C1.medium_type, sum(date_part('second', C2.date_time - C1.date_time)) as cnt
from CTE2 as C1
    left outer join CTE2 as C2 on C2.row_num = C1.row_num + 1
group by C1.medium_type

SQL FIDDLE EXAMPLE

Answer 2

我认为也可以采用r3-r2，它会给出相同的结果，不是吗？如果是这样，那么我认为这样做了。 尚未测试：

SELECT t.medium_type, t.sum(duration) FROM ( SELECT n2.date_time - n1.date_time as duration FROM networkmedium n1 INNER JOIN networkmedium n2 on n2.id = n1.id+1 ) as t GROUP BY medium_type

改变价值运行总时间

2 个答案: