如何在 MySQL 中计算 75% 的任务何时完成

Question

以电子商务为例，某些品类有一定数量的产品。如何找出每个类别中 75% 的产品售罄的时间。

例如，一天开始时，移动类别中有 20 个 Google Pixel（产品），笔记本电脑类别中有 100 个 Thinkpad，衣服类别中有 1000 个 T 恤。第 15 个像素在早上 6:35:00 售出，而第 20 个（当天最后一天）在上午 7:00:00 售出。第 75 台 Thinkpad 在下午 2:40:00 售出，而第 91 台（当天最后一天）在下午 6:30:00 售出。当天发送的第 300 个 T 恤已售出（未售出 75%）。

事务表是：

+----------+------------+-----------+
| ORDER_ID | PRODUCT_ID | SELL_TIME |
+----------+------------+-----------+
| 1        | 1          | 00:00:00  |
+----------+------------+-----------+
| 2        | 1          | 00:00:01  |
+----------+------------+-----------+
| 15       | 1          | 6:35:00   |
+----------+------------+-----------+
| 20       | 1          | 7:00:00   |
+----------+------------+-----------+
| 1        | 2          | 00:30:00  |
+----------+------------+-----------+
| 2        | 2          | 00:35:00  |
+----------+------------+-----------+
| 75       | 2          | 14:40:00  |
+----------+------------+-----------+
| 91       | 2          | 18:30:00  |
+----------+------------+-----------+
| 1        | 3          | 00:30:00  |
+----------+------------+-----------+
| 300      | 3          | 21:30:00  |
+----------+------------+-----------+

静态库存表是：

+------------+-----------+
| PRODUCT_ID | INVENTORY |
+------------+-----------+
| 1          | 20        |
+------------+-----------+
| 2          | 100       |
+------------+-----------+
| 3          | 1000      |
+------------+-----------+

答案是：

+---+---------+
| 1 |  6:35:00|
+---+---------+
| 2 | 14:40:00|
+---+---------+

Answer 1

这很棘手；）

CREATE TABLE trans(ORDER_ID int, PRODUCT_ID int , SELL_TIME date);
INSERT  trans VALUES  (1, 1, '2016-11-30'),
                    (2, 1,   '2016-11-29'),
                    (3, 1,   '2016-11-28'),
                    (4, 1,   '2016-11-27'),
                    (5, 1,   '2016-11-26'),
                    (6, 1,   '2016-11-25'),
                    (7, 2,   '2016-11-24'),
                    (8, 2,   '2016-11-23'),
                    (9, 2,   '2016-11-22'),
                    (10, 2,   '2016-11-21');


CREATE TABLE inventory(PRODUCT_ID int , INVENTORY int);
INSERT  inventory VALUES (1 ,6),
                         (2, 4);


SELECT   x.product_id, 
         Min( 
         CASE 
                  WHEN rank/inventory >= 0.75 THEN sell_time 
                  ELSE NULL 
         end) AS sell_time_75pct 
FROM     ( 
                SELECT ( 
                       CASE product_id 
                              WHEN @currproductid THEN @currow := @currow + 1 
                              ELSE @currow := 1 
                              AND    @currproductid := product_id 
                       end ) AS rank, 
                       product_id, 
                       sell_time, 
                       inventory 
                FROM   ( 
                                  SELECT     trans.product_id, 
                                             sell_time, 
                                             inventory 
                                  FROM       trans 
                                  INNER JOIN inventory 
                                  ON         ( 
                                                        inventory.product_id=trans.product_id)
                                  INNER JOIN 
                                             ( 
                                                    SELECT @currow := 0, 
                                                           @currproductid := '') r 
                                  order BY   product_id , 
                                             sell_time ASC ) x ) x 
GROUP BY product_id;

输入：

    ORDER_ID    PRODUCT_ID  SELL_TIME
1   6           1           25.11.2016 00:00:00
2   5           1           26.11.2016 00:00:00
3   4           1           27.11.2016 00:00:00
4   3           1           28.11.2016 00:00:00
5   2           1           29.11.2016 00:00:00
6   1           1           30.11.2016 00:00:00
7   10          2           21.11.2016 00:00:00
8   9           2           22.11.2016 00:00:00
9   8           2           23.11.2016 00:00:00
10  7           2           24.11.2016 00:00:00

输出：

    product_id  sell_time_75pct
1   1           29.11.2016 00:00:00
2   2           23.11.2016 00:00:00

Answer 2

这是我测试过的解决方案：

SELECT product_id, sell_time
FROM transaction 
INNER JOIN (
  SELECT product_id, MIN(order_id) AS order_id
  FROM (
    SELECT t.order_id, t.product_id,
      t.order_id >= FLOOR(i.inventory * 0.75) AS is_gt_75
    FROM transaction AS t
    INNER JOIN inventory AS i USING (product_id)
  ) AS _derived1
  WHERE is_gt_75
  GROUP BY product_id
) AS _derived2 USING (product_id, order_id);

输出：

+------------+-----------+
| product_id | sell_time |
+------------+-----------+
|          1 | 06:35:00  |
|          2 | 14:40:00  |
+------------+-----------+