PERCENTRANK (Inclusive)

[robarivas]

Excel has a PERCENTRANK function (which is different from PERCENTILE functions). I would like to find the same thing for DAX. Can anyone help me figure this out? Thank you.

[v-sihou-msft]

@robarivas

I think you just want to calculate the PERCENTERANK for current row value. Based on the logic for PERCENTRANK, we can add a RANK column in table and achieve same logic based on this column. Please see my sample below:

Create a rank column.

Rank = RANKX(Table6,Table6[Value])

Then create the PctRank based on the rank column.

PctRank = (COUNTA(Table6[Name])-Table6[Rank])/(COUNTA(Table6[Name])-1)

Regards,

[Sid10]

@v-sihou-msft Too bad PowerBI does not have the equivalent of the PercentRank function of Excel! I like your simple approach of using Rank and a computation of the Percentile. However, are you sure about your code for RankX?? It is not working when I try to duplicate your table. I get an error with your formula. Then when I try to modify it with a calculate and sum function, I only get '1' for each row!

Rank = RANKX('Table',CALCULATE(SUM('Table'[Value])))

[Anonymous]

I want to see a grouping by percentile

0-20%

20-40%

40-60%

60-80%

80-100%

based on PctRank so I can see which 'Name' fall within each percentile group.

[meezdax]

I actually answered my own question with this formula 🙂 Switch = SWITCH(TRUE(), AND(Table6[PctRank]>0,Table6[PctRank]<=0.20),"0-20%", AND(Table6[PctRank]>0.20,Table6[PctRank]<=0.40),"20-40%", AND(Table6[PctRank]>0.40,Table6[PctRank]<=0.60),"40-60%", AND(Table6[PctRank]>0.60,Table6[PctRank]<=0.80),"60-80%", (Table6[PctRank]>0.80),"80-100%")

[MattAllington]

As I understand, percentrank in Excel takes a score and returns its percentile. This seems to be exactly what percentile in DAX does.  Percentile in Excel does something different however. 

https://msdn.microsoft.com/en-us/library/dn802531.aspx

[OwenAuger · Accepted solution]

@MattAllington @robarivas

Matt - I believe the PERCENTILE functions in Excel and Power BI are similar in that they return the value sitting at a given percentile.

e.g. Excel's PERCENTILE.INC( <array>, k ) is equivalent to Power BI's PERCENTILE.INC( <column>, k ).

If k = 0.5 then they would return the 50th percentile.

PERCENTRANK in Excel is the inverse of PERCENTILE in that, for a given value (that may not appear in the array) it returns the rank expressed as a percentage.

I had a go at replicating PERCENTRANK.INC in DAX.

Sample PBIX here.

The DAX code looks like this (excessive use of variables )
Could be some bugs but works for sample data.

PercentRank.INC = 
VAR PercentRankArgument = [PercentRank Argument]
RETURN
    IF (
        // Only evaluate PercentRank for values between min/max of Number[Number] inclusive
        AND (
            PercentRankArgument >= MIN ( Number[Number] ),
            PercentRankArgument <= MAX ( Number[Number] )
        ),
        // Filter Number to values less than the PercentRankArgument
        VAR NumberLessThanArgument =
            FILTER ( Number, Number[Number] < PercentRankArgument )
        VAR NumberGreaterThanOrEqualArgument =
            FILTER ( Number, Number[Number] >= PercentRankArgument )
	// RankLower = the count of Numbers less than PercentRankArgument, and is used later for interpolation of ranks
        VAR RankLower =
            COUNTROWS ( NumberLessThanArgument )
	// NumberLower = the largest Number < PercentRankArgument, used for interpolation
        VAR NumberLower =
            MAXX ( NumberLessThanArgument, Number[Number] )
	// NumberUpper = the smallest Number >= PercentRankArgument, used for interpolation
        VAR NumberUpper =
            MINX ( NumberGreaterThanOrEqualArgument, Number[Number] )
	// PercentRankArgumentRank =  the rank of PercentRankArgument over the Number table, which is just RankLower + 1.
	// This is the same rank as NumberUpper in the Number table itself.
        VAR PercentRankArgumentRank = RankLower + 1
	// InterpolationFraction = fraction that PercentRankArgument is from NumberLower to NumberUpper  
        VAR InterpolationFraction =
            DIVIDE ( PercentRankArgument - NumberLower, NumberUpper - NumberLower )
        // Calculate the interpolated rank
	VAR RankInterpolated = RankLower
            + InterpolationFraction
            * ( PercentRankArgumentRank - RankLower )
        // Get the count of Numbers
	VAR NumberCount =
            COUNT ( Number[Number] )
        // Final PercentRank is (RankInterpolated - 1)/(NumberCount - 1)
	VAR PercentRankOutput =
            DIVIDE ( RankInterpolated - 1, NumberCount - 1 )
        RETURN
            PercentRankOutput