Question

我知道在SQL（Google Big Query）中是一个SUM函数，但是如果我需要计算间隔。例如，

create table BATTLE (battleID VARCHAR(256), countryID VARCHAR(256), yearBATTLE INT);

insert into BATTLE VALUES ('Battle1', 'Country1', 1567);
insert into BATTLE VALUES ('Battle2', 'Country2', 1568);
insert into BATTLE VALUES ('Battle3', 'Country2', 1570);
insert into BATTLE VALUES ('Battle4', 'Country3', 1599);
insert into BATTLE VALUES ('Battle5', 'Country2', 1575);
insert into BATTLE VALUES ('Battle6', 'Country2', 1620);
......

我需要找到战争的数量，其中一场战争是一个间隔小于10年的客户的战斗序列。

Answer 1

以下是BigQuery Standard SQL

#standardSQL
SELECT 
  countryID, 
  MIN(yearBATTLE) start, 
  MAX(yearBATTLE) finish,
  ARRAY_AGG(STRUCT(battleID, yearBATTLE)) battels
FROM (
  SELECT 
    battleID, countryID, yearBATTLE,
    SUM(delta) OVER(PARTITION BY countryID ORDER BY yearBATTLE) AS grp
  FROM (
    SELECT 
      battleID, countryID, yearBATTLE, 
      IF(yearBATTLE - IFNULL(LAG(yearBATTLE) OVER(PARTITION BY countryID ORDER BY yearBATTLE), yearBATTLE) <= 10, 0, 1) AS delta
    FROM `yourproject.yourdataset.battle`
  )
)
GROUP BY countryID, grp
ORDER BY countryID, grp

您可以使用您问题中的虚拟数据进行上述测试/播放（为了示例，我只是将10年改为5年）

#standardSQL
WITH `yourproject.yourdataset.battle` AS (
  SELECT 'Battle1' battleID,  'Country1' countryID, 1567 yearBATTLE UNION ALL
  SELECT 'Battle2',           'Country2',           1568 UNION ALL
  SELECT 'Battle3',           'Country2',           1570 UNION ALL
  SELECT 'Battle4',           'Country3',           1599 UNION ALL
  SELECT 'Battle5',           'Country2',           1575 UNION ALL
  SELECT 'Battle6',           'Country2',           1620 
)
SELECT 
  countryID, 
  MIN(yearBATTLE) start, 
  MAX(yearBATTLE) finish,
  ARRAY_AGG(STRUCT(battleID, yearBATTLE)) battels
FROM (
  SELECT 
    battleID, countryID, yearBATTLE,
    SUM(delta) OVER(PARTITION BY countryID ORDER BY yearBATTLE) AS grp
  FROM (
    SELECT 
      battleID, countryID, yearBATTLE, 
      IF(yearBATTLE - IFNULL(LAG(yearBATTLE) OVER(PARTITION BY countryID ORDER BY yearBATTLE), yearBATTLE) <= 5, 0, 1) AS delta
    FROM `yourproject.yourdataset.battle`
  )
)
GROUP BY countryID, grp
ORDER BY countryID, grp

结果如下

countryID   start   finish     battels.battleID    battels.yearBATTLE    
Country1    1567    1567       Battle1             1567  
Country2    1568    1575       Battle2             1568  
                               Battle3             1570  
                               Battle5             1575  
Country2    1620    1620       Battle6             1620  
Country3    1599    1599       Battle4             1599

Answer 2

你的问题不清楚，但是如果你的意思是战争是一系列的战斗，那么你可以通过将表连接到自己来获得序列：

select  *
from    BATTLE a
join    BATTLE b    on a.countryId = b.countryId
where   abs(a.yearBATTLE - b.yearBATTLE) < 10
and     a.BattleId <> b.battleId

SQL计数间隔大于N秒

2 个答案: