oracle sql中的日级计算

Question

我有两年的静态数据。我的列名为 stats_date,P2P_volume。最初我在 oracle sql developer 中创建了以下单日查询

选择“1”作为 KPI_ID，“P2P”作为 KPI_DESC，“22-MAR-17”作为日期， (sum(STATS_DATE 在 add_months('22-MAR-17',0)-13 之间的情况) 然后 add_months('22-MAR-17',0)-7 然后 P2P_VOLUME else 0 end )) LAST_WEEK_Volume,

  (sum(case when STATS_DATE between add_months('22-MAR-17',0)-6
and add_months('22-MAR-17',0)  then P2P_VOLUME else 0 end )) THIS_WEEK_Volume from table

我的问题是我想创建动态查询，这将给我 Last_week_volume 和 this_week_volume 两年的日期。而不是一天

Answer 1

在没有完整的样本数据和要求的情况下，这是我的假设：

1. 每个 KPI 每天一行
2. 可以满足current week、previous week的定义 通过使用 'IW' 日期格式掩码

此解决方案使用子查询来计算每周度量值的总和。这会输入主查询，该查询使用分析 lag() 函数来显示当前一周和前一周的总数。

with cte as (
  select kpi
        , to_char(static_date, 'YYYY') as yr
       , to_char(static_date, 'IW') as wk
       , sum(volume) as wk_volume
       , sum(value) as wk_value
       , sum(revenue) as wk_revenue
  from t23
  group by kpi,to_char(static_date, 'YYYY'), to_char(static_date, 'IW')
)
select kpi
       , yr||'-'||wk as year_wk
       , wk_volume as curr_wk_volume
       , lag(wk_volume) over (order by yr, wk) as prev_wk_volume
       , wk_value as curr_wk_value
       , lag(wk_value) over (order by yr, wk) as prev_wk_value
       , wk_revenue as curr_wk_revenue
       , lag(wk_revenue) over (order by yr, wk) as prev_wk_revenue
from cte
order by 1, 2
/

有a SQL Fiddle demo here。