- - 填充的话，那12月份的数据应该是8还是18？，如果是18就采用开窗函数组层汇总处理；如果是8，那就用「表与表的错位关联」处理，视需求自行选择；

——————————

SQL 数据分组累加sum() over (partition by ... order by ...)-https://blog.csdn.net/yole_grise/article/details/87921671

case when demand_num is null then dbo.findmonth(month) else demand_num  end

dbo.findmonth(month) 新建一个函数，根据参数月份查询上月的 demand_num 

适用于mssql

做好函数，用case when 处理为空时计算上个月的num

用窗口函数加一列为 上一行(月)  的数据，然后判断当前行是否null 取上一行(月)或取当前行

https://blog.csdn.net/rfb0204421/article/details/7675877/

有一个简单粗暴的方法，把月份加一错开和原表left join 。你把table_yourtable换成你的表试一下。

select a.usedate,

       case when a.demand_num is null then b.demand_num else a.demand_num end  as new_demand_num

from

(select CONVERT(varchar(100), create_time, 112) usedate,demand_num from table_yourtable) a

left join

(select CONVERT(varchar(100), create_time, 112)+1 usedate,demand_num from table_yourtable) b on a.usedate=b.usedate

上面是不管跨年和一月份数据，如果一月份数据也要处理就用下面语句：

select a.usedate,

       case when a.demand_num is null then b.demand_num else a.demand_num end  as new_demand_num

from

(select CONVERT(varchar(100), create_time, 112) usedate,demand_num from table_yourtable) a

left join

(select case when month(create_time) = 12 then CONVERT(varchar(100),DATEADD(yy,DATEDIFF(yy,0,dateadd(year,-1,create_time)),0), 112) 

         else CONVERT(varchar(100), create_time, 112)+1 end as usedate,

         demand_num from table_yourtable  where month(create_time)<> 01 ) b on a.usedate=b.usedate