Question

我想获取Hive中表格的第一个四分位数的摘要数据。以下是获取每个四分位数中的最大视图数的查询：

SELECT NTILE(4) OVER (ORDER BY total_views) AS quartile, MAX(total_views)
FROM view_data
GROUP BY quartile
ORDER BY quartile;

此查询是为了获取第一个四分位数中所有人的名字：

SELECT name, NTILE(4) OVER (ORDER BY total_views) AS quartile
FROM view_data
WHERE quartile = 1

我在两个查询中都收到此错误：

Invalid table alias or column reference 'quartile'

如何在ntile子句或where子句中引用group by结果？

Answer 1

您不能在window子句中放置窗口函数，因为如果存在复合谓词，它会产生歧义。所以使用子查询。

select quartile, max(total_views) from
(SELECT total_views, NTILE(4) OVER (ORDER BY total_views) AS quartile,
FROM view_data) t
GROUP BY quartile
ORDER BY quartile
;

和

select * from 
(SELECT name, NTILE(4) OVER (ORDER BY total_views) AS quartile
FROM view_data) t
WHERE quartile = 1
;

Answer 2

SQL中的WHERE语句只能在表模式中的现有列上进行选择。要在计算列上执行该功能，请使用 HAVING 而不是WHERE。

SELECT name, NTILE(4) OVER (ORDER BY total_views) AS quartile
FROM view_data
HAVING quartile = 1

使用Hive ntile导致where

2 个答案: