Sql 如何获得每列B中重复次数最多的列A

我想得到每个客户订购最多的两件商品。我可以从销售表中获得以下数据

 ---------------------------
 |OrderAccount| Item        |
 |ABC        |  Shoes #1    |
 |ABC        |  Shoes #2    |
 |ABC        |  Shoes #2    |
 |ABC        |  Shoes #1    |
 |ABC        |  Shoes #4    |
 |RDD        |  Shoes #1    |
 |RDD        |  Shoes #2    |
 |RDD        |  Shoes #1    |
 |RDD        |  Shoes #6    |
 |RDD        |  Shoes #1    |
 ----------------------------

我怎样才能得到数据呢？ 这不起作用：

SELECT so.Item,
   so.OrderAccount
  FROM (
       SELECT so.Item,
        so.OrderAccount,
        row_number() OVER(Partition BY so.Item ORDER BY so.OrderAccount desc) as repeated
       FROM SalesOrders so
     WHERE so.Item IS NOT NULL
       ) AS so
  WHERE so.repeated <= 2
  ORDER BY so.OrderAccount

---

我设法找到了一个可行的解决方案，但它使用了非常糟糕的做法，并且在实际数据库中性能很差OrderAccount重命名为customer，Item重命名为Item：

-- get the top-product per customer
SELECT customer, item, MAX(cnt) 
FROM (
  -- get all customer-item-pairs with the associated count
  SELECT customer, item, COUNT(item) AS cnt FROM tbl GROUP BY customer,item
) GROUP BY customer

UNION -- combine that with the second-top-product per customer

-- get the top-product per customer, but stripped of the first part of the result (so the second-top-product)
SELECT customer, item, MAX(cnt) 
FROM (
  -- get all customer-item-pairs with the associated count
  SELECT customer, item, COUNT(item) AS cnt FROM tbl GROUP BY customer,item
  EXCEPT --except for the customer-item-pairs which are already top-products
    --this is the same as get the top-product per customer
    SELECT customer, item, MAX(cnt) 
    FROM (
      SELECT customer, item, COUNT(item) AS cnt FROM tbl GROUP BY customer,item
    ) GROUP BY customer
) GROUP BY customer

---

这可能会起作用。如果帐户下的订单数相等，它将返回2行以上

SELECT b.OrderAccount, b.Item
FROM(
    SELECT *, RANK() OVER(PARTITION BY a.OrderAccount ORDER BY a.count_item DESC) AS RowRank
    FROM(
        SELECT so.OrderAccount, so.Item, count(item) count_item
        FROM SalesOrders so
        GROUP BY so.OrderAccount, so.Item
    ) a
) b
WHERE b.RowRank <= 2

---

哥们，你刚才犯了一个关于分区和排序的错误。您希望从每位客户那里获得前2项。所以您需要按客户进行分区，并且项目中有num，所以您需要按项目进行排序

通过：

---

当前的结果是什么？想要的结果是什么？哪个DBMS？MS SQL Server？在此查询中，订单在何处计数，以确定每个客户最常订购的项目？这是否有效？我同意下划线d，不计算发生次数。如果帐户下的订单数相等，则返回的行数将超过2行。-这是一个有益的副作用！因为op要求排名前二的Rowsorry，所以我只是想说清楚，但是如果你不得不公开承认你的答案真的很糟糕，很糟糕，那可能就不值得发布了

SELECT so.Item,
so.OrderAccount
FROM (
   SELECT so.Item,
    so.OrderAccount,
    row_number() OVER(Partition BY so.OrderAccount ORDER BY so.Item desc) as repeated
   FROM SalesOrders so
 WHERE so.Item IS NOT NULL
   ) AS so
WHERE so.repeated <= 2
ORDER BY so.OrderAccount