我想从MySql中的月度数据计算移动平均线。
我想使用每个team_id的前12个月的数据进行计算
平均销售的商品数量。
理想情况下，当特定月份遗失team_id时，我希望将该月包含为 0 。
另一个警告：
如果我们在October 2016，那么我希望移动平均窗口来自
September 2016至September 2015。
这是我的尝试不起作用：
SELECT t1.monthyear
, t2.team_id
, t1.itemssold
, AVG(t2.itemssold) AS avgTransactions
FROM t1
JOIN t1 AS t2
ON PERIOD_DIFF(t2.monthyear,t1.monthyear) = 12
AND (t2.team_id = t1.team_id)
GROUP
BY t1.team_id
, monthyear;
一些数据：
monthyear tmonth tyear team_id ItemsSold
201405 5 2014 23 1
201406 6 2014 23 3
201407 7 2014 23 5
201408 8 2014 23 3
201409 9 2014 23 7
201410 10 2014 23 1
201411 11 2014 23 4
201501 1 2015 23 6
201502 2 2015 23 2
201503 3 2015 23 2
201504 4 2015 23 5
201505 5 2015 23 5
201506 6 2015 23 6
201507 7 2015 23 11
201508 8 2015 23 13
201509 9 2015 23 5
201601 1 2016 23 14
201602 2 2016 23 4
201603 3 2016 23 3
201604 4 2016 23 10
201605 5 2016 23 9
201606 6 2016 23 7
201409 9 2014 83 1
201511 11 2015 83 1
201604 4 2016 83 4
201605 5 2016 83 1
201606 6 2016 83 3
201405 5 2014 88 1
201406 6 2014 88 3
201407 7 2014 88 4
201408 8 2014 88 3
201409 9 2014 88 6
201410 10 2014 88 7
201411 11 2014 88 5
201412 12 2014 88 11
201501 1 2015 88 8
201502 2 2015 88 5
201503 3 2015 88 6
201504 4 2015 88 3
201505 5 2015 88 11
201506 6 2015 88 6
201507 7 2015 88 12
201508 8 2015 88 14
201509 9 2015 88 11
201510 10 2015 88 12
201511 11 2015 88 14
201512 12 2015 88 13
201601 1 2016 88 7
201602 2 2016 88 10
201603 3 2016 88 15
201604 4 2016 88 11
201605 5 2016 88 14
201606 6 2016 88 10
201410 10 2014 131 2
201501 1 2015 131 1
201505 5 2015 131 1
201509 9 2015 131 2
201407 7 2014 135 2
201503 3 2015 135 1
201509 9 2015 135 1
201603 3 2016 135 1
201605 5 2016 135 1
201405 5 2014 140 15
201406 6 2014 140 18
201407 7 2014 140 14
201408 8 2014 140 21
201409 9 2014 140 22
201410 10 2014 140 19
201411 11 2014 140 18
201412 12 2014 140 13
201501 1 2015 140 25
201502 2 2015 140 17
201503 3 2015 140 26
201504 4 2015 140 17
201505 5 2015 140 14
201506 6 2015 140 16
201507 7 2015 140 20
201508 8 2015 140 17
201509 9 2015 140 23
201510 10 2015 140 19
201511 11 2015 140 12
201512 12 2015 140 4
答案 0 :(得分：0)
如果你有类似这样的日期表
+------+------------+----------+------------------+-------------------+--------+
| id | dte | CalMonth | CalMonthDescLong | CalMonthDescShort | calQtr |
+------+------------+----------+------------------+-------------------+--------+
| 1 | 2000-01-01 | 1 | January | Jan | 1 |
| 2 | 2000-01-02 | 1 | January | Jan | 1 |
| 3 | 2000-01-03 | 1 | January | Jan | 1 |
| 4 | 2000-01-04 | 1 | January | Jan | 1 |
| 5 | 2000-01-05 | 1 | January | Jan | 1 |
+------+------------+----------+------------------+-------------------+--------+
您可以为缺少的日期生成虚拟条目，例如
select distinct year(dte) * 100 + month(dte) monthyear, month(dte) as tmonth, year(dte) as tyear, t1.team_id,0 as itemssold
from dates
cross join (select distinct t.team_id from t) t1
where year(dte) * 100 + month(dte) >= (select min(monthyear) from t) and year(dte) * 100 + month(dte) <= (select max(monthyear) from t)
and t1.team_id = 23
将虚拟条目与实际数据合并，如此
select s.team_id,s.monthyear,s.tmonth,s.tyear,
sum(case when s.itemssold is null then 0 else s.itemssold end) itemsold
from
(
select distinct year(dte) * 100 + month(dte) monthyear, month(dte) as tmonth, year(dte) as tyear, t1.team_id,0 as itemssold
from dates
cross join (select distinct t.team_id from t) t1
where year(dte) * 100 + month(dte) >= (select min(monthyear) from t) and year(dte) * 100 + month(dte) <= (select max(monthyear) from t)
and t1.team_id = 23
union
select t.monthyear,t.tmonth,t.tyear,t.team_id,t.ItemsSold from t
) s
group by s.team_id,s.monthyear,s.tmonth,s.tyear
然后将其加入另一个版本并计算出移动平均线
select t2.team_id,t2.monthyear,t2.itemsold
,sum(t3.itemsold) sold,count(*) obs
,AVG(t3.itemsold * 1.00) AS avgTransactions
from
(
select s.team_id,s.monthyear,s.tmonth,s.tyear,
sum(case when s.itemssold is null then 0 else s.itemssold end) itemsold
from
(
select distinct year(dte) * 100 + month(dte) monthyear, month(dte) as tmonth, year(dte) as tyear, t1.team_id,0 as itemssold
from dates
cross join (select distinct t.team_id from t) t1
where year(dte) * 100 + month(dte) >= (select min(monthyear) from t) and year(dte) * 100 + month(dte) <= (select max(monthyear) from t)
#and t1.team_id = 23
union
select t.monthyear,t.tmonth,t.tyear,t.team_id,t.ItemsSold from t
) s
group by s.team_id,s.monthyear,s.tmonth,s.tyear
) t2
join
(
select s.team_id,s.monthyear,s.tmonth,s.tyear,
sum(case when s.itemssold is null then 0 else s.itemssold end) itemsold
from
(
select distinct year(dte) * 100 + month(dte) monthyear, month(dte) as tmonth, year(dte) as tyear, t1.team_id,0 as itemssold
from dates
cross join (select distinct t.team_id from t) t1
where year(dte) * 100 + month(dte) >= (select min(monthyear) from t) and year(dte) * 100 + month(dte) <= (select max(monthyear) from t)
#and t1.team_id = 23
union
select t.monthyear,t.tmonth,t.tyear,t.team_id,t.ItemsSold from t
) s
group by s.team_id,s.monthyear,s.tmonth,s.tyear
) t3
on t2.team_id = t3.team_id and (t3.tyear * 12 + t3.tmonth between (t2.tyear * 12 + t2.tmonth) - 11 and (t2.tyear * 12 + t2.tmonth))
group by t2.team_id,t2.monthyear desc
（部分）结果
+---------+-----------+----------+------+-----+-----------------+
| team_id | monthyear | itemsold | sold | obs | avgTransactions |
+---------+-----------+----------+------+-----+-----------------+
| 23 | 201606 | 7 | 76 | 12 | 6.333333 |
| 23 | 201605 | 9 | 75 | 12 | 6.250000 |
| 23 | 201604 | 10 | 71 | 12 | 5.916667 |
| 23 | 201603 | 3 | 66 | 12 | 5.500000 |
| 23 | 201602 | 4 | 65 | 12 | 5.416667 |
| 23 | 201601 | 14 | 63 | 12 | 5.250000 |
| 23 | 201512 | 0 | 55 | 12 | 4.583333 |
| 23 | 201511 | 0 | 55 | 12 | 4.583333 |
| 23 | 201510 | 0 | 59 | 12 | 4.916667 |
| 23 | 201509 | 5 | 60 | 12 | 5.000000 |
| 23 | 201508 | 13 | 62 | 12 | 5.166667 |
| 23 | 201507 | 11 | 52 | 12 | 4.333333 |
| 23 | 201506 | 6 | 46 | 12 | 3.833333 |
| 23 | 201505 | 5 | 43 | 12 | 3.583333 |
| 23 | 201504 | 5 | 39 | 12 | 3.250000 |
| 23 | 201503 | 2 | 34 | 11 | 3.090909 |
| 23 | 201502 | 2 | 32 | 10 | 3.200000 |
| 23 | 201501 | 6 | 30 | 9 | 3.333333 |
| 23 | 201412 | 0 | 24 | 8 | 3.000000 |
| 23 | 201411 | 4 | 24 | 7 | 3.428571 |
| 23 | 201410 | 1 | 20 | 6 | 3.333333 |
| 23 | 201409 | 7 | 19 | 5 | 3.800000 |
| 23 | 201408 | 3 | 12 | 4 | 3.000000 |
| 23 | 201407 | 5 | 9 | 3 | 3.000000 |
| 23 | 201406 | 3 | 4 | 2 | 2.000000 |
| 23 | 201405 | 1 | 1 | 1 | 1.000000 |
| 83 | 201606 | 3 | 9 | 12 | 0.750000 |
| 83 | 201605 | 1 | 6 | 12 | 0.500000 |