我有一个USE [JainPort114]
GO
/****** Object: StoredProcedure [dbo].[GetEventValveONOFF] Script Date: 17-Apr-17 9:18:14 AM ******/
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
ALTER PROCEDURE [dbo].[GetEventValveONOFF]
--@RTUID int, @StartDatetime datetime, @Enddatetime datetime,
--@networkId int, @Zoneid int, @BlockId int
@WhereCondition nvarchar(max) , @StartDatetime nvarchar(100), @Enddatetime nvarchar(100)
AS
BEGIN
SET NOCOUNT ON;
--If only 1 zone is selected
begin
DECLARE @SQL NVARCHAR(MAX)
print @StartDatetime
print @Enddatetime
SELECT @SQL = 'select be.eventdatetime,''Valve Event'' as eventtype, elest.status as ''status'' ,
r.rtuname, c.channelname,c.channelid, c.tagname, c.tagname as description,
(select d.Type from DigitalOutput d where d.DOId = c.typeid and c.subtypeid = 0 and c.EqpTypeId =4) as SensorType,
--(select Reason from JainPort114_events.dbo.ElementTypeReason where ReasonId = be.Reason and ElementTypeId = 4) as Reason,
--(select EleTypeReasonId from JainPort114_events.dbo.ElementTypeReason where ReasonId = be.Reason and ElementTypeId = 4) as ReasonId
etr.Reason, etr.ReasonId,
case when be.Reason = 8 then (Select be.Errorcode)
when be.Reason = 9 then (Select be.Errorcode)
else 0
end as moidruleid
from JainPort114_events.dbo.BSTEvents b, JainPort114_events.dbo.BSTEventsConfig be, rtu r, network n, channel c , JainPort114_events.dbo.ElementTypeReason etr,JainPort114_events.dbo.elementstatus elest
where be.bsteventid = b.BSTEventsId
and c.rtuid = r.rtuid
and r.networkid = n.networkid
and elest.EleStatusId = be.elestatus
and be.EventDateTime between '''+ @StartDatetime + ''' and ''' + @Enddatetime + '''
and b.NetworkId = (select networkno from network where NetworkId = r.networkid)
and b.NetworkId = n.networkno
and JainPort114_events.dbo.HextoDEC(SUBSTRING (be.elenumberandtype, 1, 2), 16) = r.rtuidinnw
and JainPort114_events.dbo.HextoDEC(SUBSTRING (be.elenumberandtype, 3, 2), 16) = c.slotidinrtu
and c.slotidinrtu = be.number
and etr.ElementTypeId = 4
and etr.ReasonId = be.Reason
--and n.networkid = @networkId and r.rtuid = @RTUID
--and etr.ReasonId = 13
and r.active =1'
+@WhereCondition
+'order by be.eventdatetime desc'
PRINT @SQL
EXEC sys.sp_executesql @SQL
End
END
数据框，其中有两列pandas和date（3k个不同日期，总行数800k）
我想计算按日期分组的平均值，但仅限于最低十分位数的值。
我尝试使用value，它给出了最低十分位数的DCL = df['date','value'].groupby(['date'])['value'].quantile(.1)值，对于每个日期，我如何为每个日期创建一个条件均值，以便它只使用值低于cutoff（每个日期）？
DCL每天的截止值不同，我想计算值的平均值＆＃39;按日期分组仅使用该日期的数字低于当天的截止值。
答案 0 :(得分：1)
考虑使用transform为当前行日期的十进制均值添加新列。
df['DCL'] = df[['date','value']].groupby(['date'])['value'].\
transform(lambda g: g[g <= g.quantile(.1)].mean())