问题描述
我有 2 个表,第一个表名为 Activations,有两列:Line_ID、Activation_Date。 第二个名为 Speed 的表具有以下列:Line_ID、From_Date、To_Date、Record。
第一个表格示例:
|Line_ID| Activation_Date|
|-------+----------------|
|123456 | 1-Jan |
|345678 | 2-Jan |
|987654 | 3-Jan |
...
第二个间隙和岛屿表:
|Line_ID|From_Date| To_Date |Speed|
|-------+---------+---------+-----|
|123456 |1-Jan |4-Jan |70 |
|123456 |4-Jan |7-Jan |51 |
|123456 |7-Jan |10-Jan |48 |
|123456 |10-Jan |15-Jan |40 |
|123456 |15-Jan |17-Jan |70 |
|123456 |17-Jan |19-Jan |54 |
|123456 |19-Jan |21-Jan |94 |
|123456 |21-Jan |28-Jan |91 |
|123456 |28-Jan |31-Jan |35 |
...
我需要将 Activation 表与 Records 表连接起来,以将 4 列添加到 Activation 表中,但有一些顾虑,
- 第一个:从 Activation_Date 开始的前 7 天记录的平均速度。
- 第二个:记录的第二个 7 天的平均速度。
- 3ed:记录从 Activation_Date 开始的第三个 7 天的平均速度。
- 第 4 天:记录从 Activation_Date 开始的第四个 7 天的平均速度。
结果如下
|Line_ID| Activation_Date|AVG_SPEED_Week1|AVG_SPEED_Week2|AVG_SPEED_Week3|AVG_SPEED_Week4|
|-------+----------------+---------------+---------------+---------------+---------------|
|123456 | 1-Jan |60.5 |44 |72.6 |91 |
...
结果探索
AVG_SPEED_Week1: Average of Speed in the 1st 7 days starting Records.From_Date: 1-Jan Records.To_Date: 7-Jan
AVG_SPEED_Week2: Average of Speed in the 2nd 7 days starting Records.From_Date: 8-Jan Records.To_Date: 14-Jan
AVG_SPEED_Week3: Average of Speed in the 2nd 7 days starting Records.From_Date: 15-Jan Records.To_Date: 21-Jan
AVG_SPEED_Week4: Average of Speed in the 2nd 7 days starting Records.From_Date: 22-Jan Records.To_Date: 28-Jan
解决方法
我无法测试它,但怎么样?:
SELECT a.Line_ID,a.Activation_Date,CASE WHEN a.Activation_Date >= s.From_Date AND a.Activation_Date <= s.To_Date AND DATEADD(day,-7,s.To_Date) >= a.Activation_Date THEN AVG(SUM(s.Speed)) END AVG_SPEED_Week1,-14,s.To_Date) >= a.Activation_Date AND DATEADD(day,s.From_Date) >= a.Activation_Date THEN AVG(SUM(s.Speed)) END AVG_SPEED_Week2,-21,s.From_Date) >= a.Activation_Date THEN AVG(SUM(s.Speed)) END AVG_SPEED_Week3,-28,s.From_Date) >= a.Activation_Date THEN AVG(SUM(s.Speed)) END AVG_SPEED_Week4
FROM Activations a
JOIN Speed s
ON a.Line_ID=s.Line_ID
GROUP BY a.Line_ID,a.Activation_Date
我假设您不需要动态计算和生成任意周数的平均速度,4 周就足够了。
它肯定需要测试。
,我会扩展数据并汇总:
with s as (
select s.*,s.from_date + n.idx * interval '1 day' as dte
from speed s join
_V_VECTOR_IDX n
on s.to_date <= s.from_date + n.idx * interval '1 day'
)
select a.line_id,avg(case when s.dte between a.activation_date and a.activation_date + interval '6 day' then s.speed end),avg(case when s.dte between a.activation_date + interval '7 day' and a.activation_date + interval '13 day' then s.speed end),avg(case when s.dte between a.activation_date + interval '14 day' and a.activation_date + interval '20 day' then s.speed end),avg(case when s.dte between a.activation_date + interval '21 day' and a.activation_date + interval '27 day' then s.speed end)
from activations a left join
s
on a.line_id = s.line_id
group by a.line_id,a.activation_date;
这假设时间段少于 1000 天左右。