I\'m trying to get mean, median, mode and range for a set of values in a table. I was able to get the average but median, range and mode I\'m getting a wrong one.
Below
I imagine that you may rather want to do something like this:
select dbo.Median(DataValues_AttributeValue)
from ...
There is no slick way to get the median or mode in a manner similar to using the native aggregates such as avg, max, min, max, etc. However, you may want to try a .NET CLR aggregate implementation where you implement the median and mode in, for example, C# if you want something elegant, e.g. like the above code snippet.
This is what I've done in the past.
Not sure if this will help, but here is some sql which allows me generate some stats (..., mean, median, mode,..) within a group by
I am calculated only one measure, but I suspect it can easily be expanded Where I have "GrpByYear", this would have to be expanded into your compound fields.
;with cteBase as (
Select RowNr=Row_Number() over (Partition By Year(TR_Date) Order By Year(TR_Date),TR_Y10)
,GrpByYear = Year(TR_Date)
,Measure = TR_Y10
From [Chinrus-Series].[dbo].[DS_Treasury_Rates]
Where Year(TR_Date)>2014
)
,cteMedian as (Select A.GrpByYear,Measure From cteBase A Join (Select GrpByYear,RowNr=Max(RowNr)/2 from cteBase Group by GrpByYear) B on (A.GrpByYear=B.GrpByYear and A.RowNr=B.RowNr))
,cteMode as (Select * from (Select RowNr=Row_Number() over (Partition By GrpByYear Order by Count(*) Desc),GrpByYear,Measure,Hits=count(*) From cteBase Group by GrpByYear,Measure) A Where RowNr=1)
Select A.GrpByYear
,RecordCount = Count(*)
,DistinctCount = Count(Distinct A.Measure)
,SumTotal = Sum(A.Measure)
,Minimum = Min(A.Measure)
,Maximum = Max(A.Measure)
,Mean = Avg(A.Measure)
,Median = Max(B.Measure)
,Mode = Max(C.Measure)
,StdDev = STDEV(A.Measure)
From cteBase A
Join cteMedian B on A.GrpByYear=B.GrpByYear
Join cteMode C on A.GrpByYear=C.GrpByYear
Group By A.GrpByYear
Order By A.GrpByYear
Year RecordCount DistinctCount SumTotal Minimum Maximum Mean Median Mode StdDev
2016 110 43 204.82 1.63 2.25 1.862 1.84 1.83 0.128568690811108
2015 251 69 536.71 1.68 2.50 2.1382 2.16 2.20 0.1662836533952
In SQL 2012 or later, it's often easier to use the percentile_cont
function to calculate the median. It looks like the rest of your question has already been addressed, but I thought you'd want to know about this option as well.
https://msdn.microsoft.com/en-us/library/hh231473.aspx