将缺失的日期添加到每个位置 - MySQL 5.7

时间:2021-04-21 10:15:59

标签: mysql date join calendar gaps-in-data

所以我的 SQL 输出是:

DUNF    2021-04-01  18
DUNF    2021-04-02  17
DUNF    2021-04-03  7
DUNF    2021-04-04  10
DUNF    2021-04-05  18
DUNF    2021-04-06  20
DUNF    2021-04-07  17
DUNF    2021-04-08  14
LEEDS   2021-04-01  4
LEEDS   2021-04-02  4
LEEDS   2021-04-03  5
LEEDS   2021-04-05  9
LEEDS   2021-04-06  3
LEEDS   2021-04-07  1
LEEDS   2021-04-08  3
etc.

但我需要每天都有一个条目,即使最后一个字段是一个卷(有时当天那个位置什么也没有发生)。

我什么都试过了,但我的 SQL 5.7 知识很薄弱。我可以生成一个日历表,但它并没有填补每个站点的空白(您可以看到 Leeds 缺少一天,但还有其他站点也缺少日期)。

我目前使用的代码是:

SELECT location, begin_date, count(*) as volume
  FROM abs_raw_data
  WHERE begin_date >= (SELECT date_sub(max(begin_date), interval 7 day) from abs_raw_data)
  GROUP BY location, begin_date
  ORDER BY location asc, begin_date asc

我该怎么做?

2 个答案:

答案 0 :(得分:1)

看来你需要

SELECT locations.location, 
       calendar.begin_date, 
       COALESCE(COUNT(abs_raw_data.location), 0) as volume
FROM ( {calendar generating subquery} ) AS calendar
CROSS JOIN ( SELECT DISTINCT location
             FROM abs_raw_data ) locations
CROSS JOIN ( SELECT MAX(begin_date) - INTERVAL 7 DAY AS begin_date 
             FROM abs_raw_data ) maxdate
LEFT JOIN abs_raw_data ON calendar.begin_date = abs_raw_data.begin_date 
                      AND locations.location = abs_raw_data.location
                      AND abs_raw_data.begin_date >= maxdate.begin_date 
GROUP BY locations.location, calendar.begin_date
ORDER BY locations.location ASC, calendar.begin_date ASC

日历生成子查询可以是,例如,

SELECT abs_raw_data.begin_date - INTERVAL nums.n DAY
FROM ( SELECT MAX(begin_date) begin_date
       FROM abs_raw_data) abs_raw_data 
CROSS JOIN ( SELECT 0 n UNION SELECT 1 UNION SELECT 2 UNION SELECT 3
             UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 ) nums
-- https://dbfiddle.uk/?rdbms=mysql_8.0&fiddle=49ef261eaf01956973d4ce99f6203a91

答案 1 :(得分:0)

另一个答案建议日期的生成。但是,我会创建一个名为 years 的表

years(year)

另一个叫做月如

months(month)

和另一个称为天

days(month, day)

正确填写所有内容,包括 2 月 29 日。因此,如果您想以这种方式生成日期和月份,那么您将需要一个连接,例如:

select years.year, months.month, days.day
from years
join months
on 1
join days
on months.month = days.month and
   ((days.day <> 29) or (months.month <> 2) or (years.year % 4 > 0))
where (
      years.year > 2020 and
      years.year < 2025 and
      ) or 
      (
      years.year = 2020 and
      years.year = 2025 and
      months.month < 11 and
      months.month > 2 and
      ) or
      (
      years.year = 2020 and
      years.year = 2025 and
      months.month = 11 and
      months.month = 2 and
      days.day >= 2 and
      days.day <= 21 and
      )

确保插入您可能使用的任何年份,插入一年中的所有月份和一个月中的所有日期,然后将上述内容用作子查询。