Dynamically Exclude 10 Percent of an Asset's Highest Values

[Anonymous]

Hi all,

I am attempting to create a dynamic measure that returns an asset’s highest value after 10% of the asset’s highest values are removed.

For instance, I have a table like the one below. Asset A1 has 20 rows associated with it, while A2 has 10. I would like my measure to return the 18th value for A1 (188) and the 9th value for A2 (193).

             Unordered                                                   Ordered

 

Currently, I am attempting to countrows then, using the TOPN and EXCPET functions, return a temporary table where the MAX value is the number I’m trying to get at. I'm missing something because my measure doesn't return anything. 

My current code looks like this:

90PercentofTime =

VAR sampleSize = COUNTROWS(SampleData)

VAR tenPercent = .10

RETURN

CALCULATE(

    MAX(SampleData[Value]),

    EXCEPT(

        TOPN(sampleSize*tenPercent, SampleData), SampleData))

Any help would be greatly appreciated.

Thank you!

[LivioLanzo · Accepted solution]

Hi @

ttry like this:

Measure = 
IF(
    HASONEVALUE( SampleData[AssetID] ),
    VAR Top10Perc = PERCENTILE.INC(SampleData[Value], 0.9 )
    RETURN
    CALCULATE(
        MAX( SampleData[Value] ),
        SampleData[Value] < Top10Perc
    )
)

[Anonymous]

I think that got it. Thank you!

[Omega]

Try: 

Measure = calculate (max (value), allexcept (table1, AssetID))

[Anonymous]

Hi Omega, thanks for the suggestion. I think this is in the right direction, but I still need to know how many of an asset's highest values to exclude to equal 10% of its data (rows). Maybe adding TOPN in before or after MAX?

[Omega]

If I understood this correctly, basically you are trying to apply the following: 

1. Find the max value at assetID level
2. How much I need to decrease from this asset to equal 10%

Right?

[Anonymous]

That's correct. For instance, in the table above, asset A1 has 20 rows of data associated with it, so 10% of A1's rows would equal 2. Therefore I want to exlude the top 2 highest values from A1's data so that I'm left with 188 as A1's highest value. The same applied to A2, however since there are only 10 rows of data, I would only exlude 1 row. The highest value for A2 would then be 193. Does that make sense?