Select 每組的第一行和最后一行並取 MySQL 中的列值差異？

Question

我有以下表結構

象征	日期	價格
蘋果	2021-02-21	10
特斯拉	2021-02-21	800
亞馬遜	2021-02-21	90
蘋果	2021-02-20	17
特斯拉	2021-02-20	900
亞馬遜	2021-02-20	105
蘋果	2021-02-19	5
特斯拉	2021-02-19	960
亞馬遜	2021-02-19	80
……	……	……

問題：現在，對於給定的日期范圍date >= '2021-02-19' and date <= '2021-02-21'我希望找到該日期范圍的價格差異。 也就是執行查詢后，下面是我想要的output，

象征	PRICE_CHANGE
蘋果	5
特斯拉	-160
亞馬遜	10
……	……

我的嘗試：

(SELECT symbol, "date"
FROM public.ticks 
where "date" >= '2021-02-19 11:30:00' and  "date" <= '2021-02-19 12:30:00'
order by "date" desc) t group by t.symbol;

我無法弄清楚如何通過一個組來確定 go 並選擇該組的第一行和最后一行並減去價格。

任何幫助是極大的贊賞！

Answer 1

使用 MySQL-8.0/MariaDB-10.2+ window function：

SELECT symbol,
       LAST - FIRST AS price_change
FROM
  (SELECT DISTINCT symbol,
                   first_value(price) OVER w AS FIRST,
                   last_value(price) OVER w AS LAST
   FROM ticks WINDOW w AS (PARTITION BY symbol
                           ORDER BY date
                           ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING)
   ) AS p

參考： 小提琴

Answer 2

我至少能想到幾個想法。

第一個想法：

SELECT symbol, 
       SUBSTRING_INDEX(prices,',',1)-SUBSTRING_INDEX(prices,',',-1) AS price_change
FROM
    (SELECT symbol, GROUP_CONCAT(price ORDER BY `date` DESC) prices
      FROM ticks 
      WHERE `date` >= '2021-02-19' AND `date` <= '2021-02-21'
 GROUP BY symbol) a;

第二個想法：

SELECT a.symbol, 
       SUM(CASE WHEN a.`date`=max_date THEN price END) AS u,
       SUM(CASE WHEN a.`date`=min_date THEN price END) AS m,
       SUM(CASE WHEN a.`date`=max_date THEN price END)-SUM(CASE WHEN a.`date`=min_date THEN price END) AS price_change
FROM ticks a 
JOIN 
   (SELECT symbol, MIN(`date`) AS min_date, MAX(`date`) AS max_date
      FROM ticks WHERE `date` >= '2021-02-19' AND `date` <= '2021-02-21'
 GROUP BY symbol) b
 ON a.symbol=b.symbol
 GROUP BY a.symbol;

小提琴演示： https://dbfiddle.uk/?rdbms=mysql_5.7&fiddle=0168329820ddbacfa5d0356a6f3ba2ea

Answer 3

我們可以使用臨時表來計算結果。 所有現代 MySQL 服務器版本都支持臨時表。 它們僅在當前連接期間持續。 有關詳細信息，請參閱臨時表的描述。

請參閱以下 SQL：

CREATE TEMPORARY TABLE min_date_tbl (
    `symbol` varchar(4),
    `min_date` datetime    
);

CREATE TEMPORARY TABLE max_date_tbl (
    `symbol` varchar(4),
    `max_date` datetime    
);

CREATE TEMPORARY TABLE min_price_tbl (
    `symbol` varchar(4),
    `min_price` int    
);

CREATE TEMPORARY TABLE max_price_tbl (
    `symbol` varchar(4),
    `max_price` int    
);

INSERT INTO min_date_tbl (SELECT symbol, MIN(date) as min_date from ticks where date >= '2021-02-19' and date <= '2021-02-21' group by symbol);

INSERT INTO max_date_tbl (SELECT symbol, MAX(date) as max_date from ticks where date >= '2021-02-19' and date <= '2021-02-21' group by symbol);

INSERT INTO min_price_tbl (SELECT t1.symbol, t1.price as min_price FROM ticks t1, min_date_tbl t2 WHERE t1.symbol=t2.symbol AND t1.date=t2.min_date);

INSERT INTO max_price_tbl (SELECT t1.symbol, t1.price as max_price FROM ticks t1, max_date_tbl t2 WHERE t1.symbol=t2.symbol AND t1.date=t2.max_date);


SELECT t1.symbol as SYMBOL, (t2.max_price - t1.min_price) as PRICE_CHANGE FROM min_price_tbl t1, max_price_tbl t2 WHERE t1.symbol=t2.symbol

我在dbfiddle 上創建並測試了這個 SQL 代碼。 它產生了所需的結果。