计算多个时间戳之间的差异

Posted 2023-03-31

【中文标题】计算多个时间戳之间的差异

【英文标题】：Calculate difference between multiple timestamps

【发布时间】：2020-10-19 13:18:23

【问题描述】：

DB-Fiddle:

CREATE TABLE sales (
    id int auto_increment primary key,
    time_stamp DATE,
    product VARCHAR(255),
    sales_quantity INT
);

INSERT INTO sales
(time_stamp, product, sales_quantity
)
VALUES 
("2020-01-14", "Product_A", "100"),
("2020-01-14", "Product_B", "300"),
("2020-01-14", "Product_C", "600"),

("2020-01-15", "Product_A", "100"),
("2020-01-15", "Product_B", "350"),
("2020-01-15", "Product_C", "600"),

("2020-01-16", "Product_A", "130"),
("2020-01-16", "Product_B", "350"),
("2020-01-16", "Product_C", "670"),
("2020-01-16", "Product_D", "400"),

("2020-01-17", "Product_A", "130"),
("2020-01-17", "Product_B", "350"),
("2020-01-17", "Product_C", "700"),
("2020-01-17", "Product_D", "450");

预期结果：

time_stamp          difference
2020-01-14             0   
2020-01-15            50   = (100+350+600) - (100+300+600)
2020-01-16           500   = (130+350+670+400) - (100+350+600)
2020-01-17            80   = (130+350+700+450) - (130+350+670+400)

---

在上面的结果中，我想计算多个timestamps 之间的sales_quantiy 的difference。

我试着用这个查询：

SELECT
time_stamp,
(SUM(t1.time_stamp_01) - SUM(t1.time_stamp_02)) AS difference
FROM
  (SELECT 
  time_stamp,
  SUM(CASE WHEN time_stamp  = '2020-01-14' THEN sales_quantity ELSE 0 END) AS time_stamp_01,
  SUM(CASE WHEN time_stamp  = '2020-01-15' THEN sales_quantity ELSE 0 END) AS time_stamp_02
  FROM sales
  GROUP BY 1) t1
GROUP BY 1;

但是，它不知何故不会计算timestamps 之间的difference，它也仅限于两个timestamps。

我需要如何修改query才能得到预期的结果？

【参考方案1】：

如果你运行的是 mysql 8.0，你可以使用聚合和lag():

select time_stamp, 
    sum(sales_quantity) 
        - lag(sum(sales_quantity), 1, sum(sales_quantity)) over(order by time_stamp) diff
from sales
group by time_stamp
order by time_stamp

这适用于 MySQL 8.0，但不适用于 MariaDB 10.3（这是您的小提琴使用的）。您可以使用：

select time_stamp, sum_sales_quantity - coalesce(lag(sum_sales_quantity) over(order by time_stamp), sum_sales_quantity) diff
from (
    select time_stamp, sum(sales_quantity) sum_sales_quantity
    from sales
    group by time_stamp
) t
order by time_stamp