Question

我有一个子查询，它连接我的客户和交易表，别名为jq。我想按订单时间戳（order_ts）创建每个客户的购买（交易）排名。所以我做到了，

SELECT customer_id, 
       order_id, 
       order_ts, 
       RANK() OVER (PARTITION BY customer_id ORDER BY order_ts ASC) AS purchase_rank, 
       amount 
FROM jq GROUP BY customer_id 
        ORDER BY customer_id;

除purchase_rank列外，我还想了解客户已完成的总购买量。所以这就变成了：

SELECT customer_id, 
       order_id, 
       order_ts, 
       RANK() OVER (PARTITION BY customer_id ORDER BY order_ts ASC) AS purchase_rank, 
       // total purchases of this customer, counted by order_id
       amount 
FROM jq GROUP BY customer_id 
        ORDER BY customer_id;

有些order_id是重复的，所以我想明确计算。如何在不加入子查询的情况下在MS SQL Server和Postgres中执行此操作？

Answer 1

Microsoft SQL Server

SELECT  customer_id,
    order_id,
    order_ts,
    purchase_rank,
    MAX(cnt) OVER(PARTITION BY customer_id) AS purchase_cnt,
    amount
FROM
(
    SELECT customer_id, 
           order_id, 
           order_ts, 
           RANK() OVER (PARTITION BY customer_id ORDER BY order_ts ASC) AS purchase_rank, 
           DENSE_RANK() OVER (PARTITION BY customer_id ORDER BY order_id ASC) AS cnt, 
           amount 
    FROM jq 
    -- GROUP BY customer_id 
) AS D  
ORDER BY customer_id;

抱歉，我不熟悉postgresql。

同时使用RANK与COUNT（DISTINCT）聚合

1 个答案:

Microsoft SQL Server