计算一段时间内不同 ID 的平均数量（1 个月、3 个月、6 个月、9 个月和 12 个月）
[英]Calculate average number of distinct ID's in a period(1 month,3 months ,6 months, 9 months and 12 months)
问题描述
Suppose I have the following source table(I filled in data only for year 2017 in order to save some space, but you can imagine, that table has data from 2017-2021)
假设我有以下源表（为了节省空间，我只填写了 2017 年的数据，但您可以想象，该表有 2017-2021 年的数据）
EMPL_ID TIMESTAMP PART_COL
1 2017-01-01 00:00:00 M
2 2017-01-01 00:00:00 M
3 2017-01-01 00:00:00 M
3 2017-01-01 00:00:00 M
1 2017-02-01 00:00:00 M
2 2017-02-01 00:00:00 M
3 2017-02-01 00:00:00 M
3 2017-02-01 00:00:00 M
1 2017-03-01 00:00:00 M
2 2017-03-01 00:00:00 M
3 2017-03-01 00:00:00 M
1 2017-04-01 00:00:00 M
2 2017-04-01 00:00:00 M
3 2017-04-01 00:00:00 M
1 2017-05-01 00:00:00 M
2 2017-05-01 00:00:00 M
3 2017-05-01 00:00:00 M
4 2017-05-01 00:00:00 M
5 2017-05-01 00:00:00 M
1 2017-06-01 00:00:00 M
2 2017-06-01 00:00:00 M
3 2017-06-01 00:00:00 M
4 2017-06-01 00:00:00 M
4 2017-06-01 00:00:00 M
1 2017-07-01 00:00:00 M
2 2017-07-01 00:00:00 M
3 2017-07-01 00:00:00 M
1 2017-08-01 00:00:00 M
2 2017-08-01 00:00:00 M
3 2017-08-01 00:00:00 M
1 2017-09-01 00:00:00 M
2 2017-09-01 00:00:00 M
1 2017-10-01 00:00:00 M
2 2017-10-01 00:00:00 M
2 2017-10-01 00:00:00 M
3 2017-10-01 00:00:00 M
4 2017-10-01 00:00:00 M
1 2017-11-01 00:00:00 M
2 2017-11-01 00:00:00 M
2 2017-11-01 00:00:00 M
3 2017-11-01 00:00:00 M
1 2017-12-01 00:00:00 M
2 2017-12-01 00:00:00 M
2 2017-12-01 00:00:00 M
3 2017-12-01 00:00:00 M
I want to calculate the following things:我想计算以下内容：
- For january I want to calculate 4 things:一月份我想计算四件事：
- number of unuqie empl_id for january partitioned by PART_COL PART_COL 分区的 1 月的 unuqie empl_id 数
- average number of unique empl_id over 3 months(from january until march) partitioned by PART_COL PART_COL 划分的 3 个月内（从 1 月到 3 月）唯一 empl_id 的平均数量
- average number of unique empl_id over 6 months(from january until june) partitioned by PART_COL PART_COL 划分的 6 个月内（从 1 月到 6 月）唯一 empl_id 的平均数量
- average number of unique empl_id over 9 months(from january until september) partitioned by PART_COL由 PART_COL 划分的 9 个月内（从 1 月到 9 月）的唯一 empl_id 的平均数
- average number of unique empl_id over 12 months(from january until december) partitioned by PART_COL PART_COL 划分的 12 个月内（从 1 月到 12 月）唯一 empl_id 的平均数量
- For the rest of the months i want to calculate在剩下的几个月里，我想计算
- number of unuqie empl_id for a month partitioned by PART_COL按 PART_COL 划分的一个月的 unuqie empl_id 数
At the end it should look like this:最后它应该是这样的：
UNIQUE_EMPL_ID TIMESTAMP_FROM TIMESTAMP_UNTIL PART_COL
3,00 2017-01-01 00:00:00 2017-01-01 00:00:00 M
3,00 2017-01-01 00:00:00 2017-03-01 00:00:00 M
3,00 2017-01-01 00:00:00 2017-06-01 00:00:00 M
3,50 2017-01-01 00:00:00 2017-09-01 00:00:00 M
3,20 2017-01-01 00:00:00 2017-12-01 00:00:00 M
3,08 2017-02-01 00:00:00 2017-02-01 00:00:00 M
3,00 2017-03-01 00:00:00 2017-03-01 00:00:00 M
3,00 2017-04-01 00:00:00 2017-04-01 00:00:00 M
5,00 2017-05-01 00:00:00 2017-05-01 00:00:00 M
and so until 12 month.直到 12 个月。
I have come up with the following query:我提出了以下查询：
SELECT
count(distinct empl_id ) as UNIQUE_EMPL_ID
,TIMESTAMP_COL as TIMESTAMP_FROM
,TIMESTAMP_COL as TIMESTAMP_UNTIL
from source_table
WHERE PART_COLUMN = 'M'
group by TIMESTAMP_COL,PART_COLUMN
union all
SELECT
count(distinct empl_id ) as UNIQUE_EMPL_ID
,TIMESTAMP_COL as TIMESTAMP_FROM
,add_months(TIMESTAMP_COL,2) as TIMESTAMP_UNTIL
,avg(count( distinct empl_id)) OVER (PARTITION BY PART_COLUMN ORDER BY TIMESTAMP_COL ROWS BETWEEN CURRENT ROW AND 2 FOLLOWING) as UNIQUE_EMPL_ID
from source_table
WHERE PART_COLUMN = 'M'
group by TIMESTAMP_COL,PART_COLUMN
union all
SELECT
count(distinct empl_id ) as UNIQUE_EMPL_ID
,TIMESTAMP_COL as TIMESTAMP_FROM
,add_months(TIMESTAMP_COL,5) as TIMESTAMP_UNTIL
,avg(count( distinct empl_id)) OVER (PARTITION BY PART_COLUMN ORDER BY TIMESTAMP_COL ROWS BETWEEN CURRENT ROW AND 5 FOLLOWING) as UNIQUE_EMPL_ID
from source_table
WHERE PART_COLUMN = 'M'
group by TIMESTAMP_COL,PART_COLUMN
union all
SELECT
count(distinct empl_id ) as UNIQUE_EMPL_ID
,TIMESTAMP_COL as TIMESTAMP_FROM
,add_months(TIMESTAMP_COL,8) as TIMESTAMP_UNTIL
,avg(count( distinct empl_id)) OVER (PARTITION BY PART_COLUMN ORDER BY TIMESTAMP_COL ROWS BETWEEN CURRENT ROW AND 8 FOLLOWING) as UNIQUE_EMPL_ID
from source_table
WHERE PART_COLUMN = 'M'
group by TIMESTAMP_COL,PART_COLUMN
union all
SELECT
count(distinct empl_id ) as UNIQUE_EMPL_ID
,TIMESTAMP_COL as TIMESTAMP_FROM
,add_months(TIMESTAMP_COL,11) as TIMESTAMP_UNTIL
,avg(count( distinct empl_id)) OVER (PARTITION BY PART_COLUMN ORDER BY TIMESTAMP_COL ROWS BETWEEN CURRENT ROW AND 11 FOLLOWING) as UNIQUE_EMPL_ID
from source_table
WHERE PART_COLUMN = 'M'
group by TIMESTAMP_COL,PART_COLUMN
Question is: can this result be achieved by more efficient query?问题是：这个结果可以通过更高效的查询来实现吗？
1 个解决方案
解决方案1
3 已采纳 2021-07-12 16:10:25
By first creating a table using a Common Table Expression, you can then use the data there to use the UNPIVOT function to calculate the totals for each period, then a simple select from the CTE to get your monthly data.首先使用通用表表达式创建一个表，然后您可以使用那里的数据使用UNPIVOT函数来计算每个时期的总数，然后从 CTE 中进行简单的选择以获得您的每月数据。
Any filters (WHERE clauses) to limit the data that needs to be returned should be added to the unique_emps CTE below.任何限制需要返回的数据的过滤器（WHERE 子句）都应该添加到下面的unique_emps CTE 中。
WITH
sample_data (emp_id, timestamp, part_col)
AS
(SELECT 1, DATE '2017-01-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-01-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-01-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-01-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-02-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-02-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-02-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-02-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-03-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-03-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-03-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-04-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-04-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-04-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-05-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-05-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-05-01', 'M' FROM DUAL
UNION ALL
SELECT 4, DATE '2017-05-01', 'M' FROM DUAL
UNION ALL
SELECT 5, DATE '2017-05-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-06-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-06-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-06-01', 'M' FROM DUAL
UNION ALL
SELECT 4, DATE '2017-06-01', 'M' FROM DUAL
UNION ALL
SELECT 4, DATE '2017-06-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-07-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-07-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-07-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-08-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-08-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-08-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-09-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-09-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-10-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-10-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-10-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-10-01', 'M' FROM DUAL
UNION ALL
SELECT 4, DATE '2017-10-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-11-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-11-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-11-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-11-01', 'M' FROM DUAL
UNION ALL
SELECT 1, DATE '2017-12-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-12-01', 'M' FROM DUAL
UNION ALL
SELECT 2, DATE '2017-12-01', 'M' FROM DUAL
UNION ALL
SELECT 3, DATE '2017-12-01', 'M' FROM DUAL),
--Query starts here
--Filters should be added to the unique_emps Common Table Expression to limit data returned
unique_emps (mon, part_col, distinct_emps)
AS
( SELECT timestamp, part_col, COUNT (DISTINCT emp_id)
FROM sample_data
GROUP BY timestamp, part_col)
SELECT timestamp_from,
ADD_MONTHS (timestamp_from, period) AS timestamp_until,
AVG (period_people) AS avg_number_of_people,
part_col
FROM (SELECT TRUNC (mon, 'Y')
AS timestamp_from,
part_col,
CASE WHEN TO_NUMBER (TO_CHAR (mon, 'Q')) <= 1 THEN distinct_emps END
period_people1,
CASE WHEN TO_NUMBER (TO_CHAR (mon, 'Q')) <= 2 THEN distinct_emps END
period_people2,
CASE WHEN TO_NUMBER (TO_CHAR (mon, 'Q')) <= 3 THEN distinct_emps END
period_people3,
CASE WHEN TO_NUMBER (TO_CHAR (mon, 'Q')) <= 4 THEN distinct_emps END
period_people4
FROM unique_emps)
UNPIVOT (period_people
FOR period
IN (period_people1 AS 2,
period_people2 AS 5,
period_people3 AS 8,
period_people4 AS 11))
GROUP BY timestamp_from, period, part_col
UNION ALL
SELECT mon, mon, distinct_emps, part_col FROM unique_emps
ORDER BY timestamp_from, timestamp_until, avg_number_of_people;
Result结果
TIMESTAMP_FROM TIMESTAMP_UNTIL AVG_NUMBER_OF_PEOPLE PART_COL
_________________ __________________ ___________________________________________ ___________
01-JAN-17 01-JAN-17 3 M
01-JAN-17 01-MAR-17 3 M
01-JAN-17 01-JUN-17 3.5 M
01-JAN-17 01-SEP-17 3.22222222222222222222222222222222222222 M
01-JAN-17 01-DEC-17 3.25 M
01-FEB-17 01-FEB-17 3 M
01-MAR-17 01-MAR-17 3 M
01-APR-17 01-APR-17 3 M
01-MAY-17 01-MAY-17 5 M
01-JUN-17 01-JUN-17 4 M
01-JUL-17 01-JUL-17 3 M
01-AUG-17 01-AUG-17 3 M
01-SEP-17 01-SEP-17 2 M
01-OCT-17 01-OCT-17 4 M
01-NOV-17 01-NOV-17 3 M
01-DEC-17 01-DEC-17 3 M
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.
相关问题
计算每个月过去 12 个月的平均值？ - Calculate the average of the last 12 months for every month?
计算过去 12 个月每月的活跃客户数量 - Calculate number of active customers each month for last 12 months
3个月内不同的累计计数 - Cumulative count distinct for a period of 3 months
参数月份选择：查询显示上个月，12个月前和最近12个月的平均值 - Parameter month selection: Make a query that shows previous month, 12 months ago and last 12 months average
缺少月份的 3 个月滚动平均值 - 3 month rolling average with missing months
计算所有12个月的GrandTotal - Calculate GrandTotal of all 12 months
连续12个月（每个月回溯11个月）时每位医生每月的不同患者计数 - count of distinct patients per doctor per month for a rolling period of 12 month(11 months look back from the current month)
MySQL查询按年龄，3个月，6个月，9个月和12个月分组 - Mysql query to group by age, 3 months, 6 months, 9 months and 12 months
(Teradata) 获取前 12 个月、3 个月和 1 个月的 SUM 值 - (Teradata) Get a SUM value of previous 12 months, 3 months and 1 month
SQL 查找过去 12 个月（不包括某些月份）的最大连续月份数 - SQL Find max no of consecutive months over a period of last 12 Months excluding certain months
相关标签