需要按总金额检索前 3 个供应商，每个类别 - 子类别

Question

我有这个数据库表：http://sqlfiddle.com/#!9/e1dd38e/7

我正在尝试查询它，以便它遍历每个类别，返回每个子类别的前 3 名供应商，按他们的总支出（总和（金额））排序，前 3 名之外的供应商将被汇总为总数并被包括在内作为“其他人”

还包括类别和子类别的总额

我很感激任何帮助，在这方面苦苦挣扎

预期结果将是：

Paper 515.54
    Napkins 314.99
        Paper LLC 110.49
        Dunder Mifflin 70.99
        Brawny 60.02
        Others 73.49
    Towels 200.55
        Acme LLC 52.01
        Michigan Paper 41.23
        Cascades LLC 40.55
        Others 66.76
Food 205.54
    Apples 178.76
        Orchards LLC 97.10
        Northstar Orchard 41.23
        Apples LLC 40.43
    Pizza 26.78
        Dominos 16.53
        Pizza Hut 10.25

Answer 1

您可以在ROW_NUMBER() 函数的帮助下使用两个级别的聚合来定义其他供应商的唯一组，这些供应商的总和排名为 4 或更多，如下所示：

SELECT category, subcategory, 
       CASE 
         WHEN grp <= 3 THEN MAX(vendor) 
         ELSE 'Others' 
       END AS vendor, 
       SUM(sum_amount) AS total_amount
FROM
(
  SELECT category, subcategory, vendor, SUM(amount) sum_amount,
       CASE 
         WHEN ROW_NUMBER() OVER (PARTITION BY category, subcategory ORDER BY SUM(amount) DESC) <=3
         THEN ROW_NUMBER() OVER (PARTITION BY category, subcategory ORDER BY SUM(amount) DESC)
         ELSE 4
       END AS grp
  FROM purchases
  GROUP BY category, subcategory, vendor
) T
GROUP BY category, subcategory, grp
ORDER BY category DESC, subcategory, grp 

请参阅demo。

如果可能有多个供应商具有相同的前三个金额总和，并且您希望将它们全部列出（不仅仅是其中三个），则将 ROW_NUMBER() 函数替换为 DENSE_RANK() 函数，如下所示：

SELECT category, subcategory, 
       grp AS vendor,
       SUM(sum_amount) AS total_amount
FROM
(
  SELECT category, subcategory, vendor, SUM(amount) sum_amount,
       CASE 
         WHEN dense_rank() OVER (PARTITION BY category, subcategory ORDER BY SUM(amount) DESC) <=3
         THEN vendor
         ELSE 'Others'
       END AS grp
  FROM purchases
  GROUP BY category, subcategory, vendor
) T
GROUP BY category, subcategory, grp
ORDER BY category DESC, subcategory, CASE WHEN grp<>'Others'THEN 1 ELSE 2 END, SUM(sum_amount) DESC 

在修改后的数据集上查看demo。