如何通过mysql获取组中的第一条，最大，最小，最后一条记录

Question

can anyone help me to know first record from open column, max of from high, min of from low, and last record from close column with 5 minutes interval format group

https://drive.google.com/open?id=1HMDECnuReJbnmRj0o_Gevn1ePMdpDoNy

I tried the below query for interval wise group by

SELECT
    timestamp,
    symbol,
    open,
    MAX(high) as high,
    MIN(low) as low,
    close
FROM ohlc_database 
WHERE `symbol` LIKE 'AMBUJACEM' 
GROUP BY 
UNIX_TIMESTAMP(timestamp) DIV 300, symbol

Expected Output

Answer 1

One of the tricks to get the first & last values in MySql 5.x is to combine GROUP_CONCAT (with an order by) & SUBSTRING_INDEX .

Example:

SELECT 
 symbol,
 FROM_UNIXTIME(FLOOR(UNIX_TIMESTAMP (`timestamp`)/300)*300) AS `timestamp`,
 CAST(SUBSTRING_INDEX(GROUP_CONCAT(`open` ORDER BY `timestamp` ASC SEPARATOR '|'),'|',1) AS DECIMAL(10,2)) AS `open`,
 MAX(`high`) AS `high`,
 MIN(`low`) AS `low`,
 CAST(SUBSTRING_INDEX(GROUP_CONCAT(`close` ORDER BY `timestamp` DESC SEPARATOR '|'),'|',1) AS DECIMAL(10,2)) AS `close`
FROM ohlc_database
WHERE symbol LIKE 'AMBUJACEM'
GROUP BY symbol, FROM_UNIXTIME(FLOOR(UNIX_TIMESTAMP (`timestamp`)/300)*300);

Result:

symbol      timestamp           open    high    low     close
---------   ------------------- ------  ------  ------  ------
AMBUJACEM   24.01.2019 03:45:00 213,10  213,5   213,1   213,50
AMBUJACEM   24.01.2019 03:50:00 213,70  213,8   212     212,40
AMBUJACEM   24.01.2019 03:55:00 212,40  212,75  211,85  211,90

In MySql 8, you could also use the window function FIRST_VALUE for this.

SELECT 
 symbol, 
 ts05 as `timestamp`, 
 MIN(first_open) as `open`,
 MAX(`high`) as `high`,
 MIN(`low`) as `low`,
 MAX(last_close) as `close`
FROM
(
  SELECT 
   id, symbol, `timestamp`, `low`, `high`, 
   FROM_UNIXTIME(FLOOR(UNIX_TIMESTAMP (`timestamp`)/300)*300) AS ts05,
   FIRST_VALUE(`open`) OVER (PARTITION BY symbol, FROM_UNIXTIME(FLOOR(UNIX_TIMESTAMP (`timestamp`)/300)*300) ORDER BY `timestamp` ASC) as first_open,
   FIRST_VALUE(`close`) OVER (PARTITION BY symbol, FROM_UNIXTIME(FLOOR(UNIX_TIMESTAMP (`timestamp`)/300)*300) ORDER BY `timestamp` DESC) AS last_close
  FROM ohlc_database
  WHERE symbol LIKE 'AMBUJACEM'
) q
GROUP BY symbol, ts05;

A test on db<>fiddle here

Answer 2

The posted dataset appears to be different. Anyway, does this help:

SELECT FROM_UNIXTIME(FLOOR(UNIX_TIMESTAMP(`timestamp`)/300)*300) x FROM ohlc_database;
+---------------------+
| x                   |
+---------------------+
| 2019-01-24 03:45:00 |
| 2019-01-24 03:45:00 |
| 2019-01-24 03:45:00 |
| 2019-01-24 03:45:00 |
| 2019-01-24 03:45:00 |
| 2019-01-24 03:50:00 |
| 2019-01-24 03:50:00 |
| 2019-01-24 03:50:00 |
| 2019-01-24 03:50:00 |
| 2019-01-24 03:50:00 |
| 2019-01-24 03:55:00 |
| 2019-01-24 03:55:00 |
| 2019-01-24 03:55:00 |
| 2019-01-24 03:55:00 |
| 2019-01-24 03:55:00 |
+---------------------+

Answer 3

Use a subquery to get the minimum and maximum timestamps for each each period and symbol. Then use conditional aggregation to capture the values at those timestamp.

SELECT UNIX_TIMESTAMP(timestamp) DIV 300 as ts,
       symbol,
       MAX(CASE WHEN timestamp = min_ts THEN open END) as open,
       MAX(high) as high,
       MIN(low) as low,
       MAX(CASE WHEN timestamp = max_ts THEN close END) as close
FROM ohlc_database o JOIN
     (SELECT symbol, 
             UNIX_TIMESTAMP(timestamp) DIV 300 as ts,
             MIN(timestamp) as min_ts, MAX(timestamp) as max_ts
      FROM ohlc_database
      WHERE symbol LIKE 'AMBUJACEM' 
      GROUP BY symbol, UNIX_TIMESTAMP(timestamp) DIV 300
     ) o2
     ON o2.ts = UNIX_TIMESTAMP(o.timestamp) DIV 300 AND
        o2.symbol = o.symbol
WHERE symbol LIKE 'AMBUJACEM' 
GROUP BY UNIX_TIMESTAMP(timestamp) DIV 300, symbol