Forum Discussion

smpa01's avatar
smpa01
Community Champion
3 years ago
Solved

LINEST forEach

AlexisOlson bcdobbs CNENFRNL    I am trying to crerate a measure using LINEST that would give me the slope but I have not been successful.   PFA what I have tried.   Following is what I d...
Need Help
  • v-jianboli-msft's avatar
    v-jianboli-msft
    3 years ago

    Hi smpa01 ,

     

    Please try:

    Measure = 
var _a = FILTER(ALL('fact'),'fact'[Country]=MAX('fact'[Country]))
return MAXX(CALCULATETABLE(LINEST('fact'[y],'fact'[x]),_a),[Slope1])

    Final output:

    Best Regards,

    Jianbo Li

    If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

  • AlexisOlson's avatar
    AlexisOlson
    3 years ago

    smpa01 You could create a measure that returns a text table using TOCSV and parse that.

     

    For example:

    LINEST Output = 
VAR _filteredTable_ = FILTER ( ALL ( 'fact' ), 'fact'[Country] = MAX ( 'fact'[Country] ) )
VAR _linEst_ = LINESTX ( _filteredTable_, 'fact'[y], 'fact'[x] )
VAR _toText = TOCSV ( _linEst_, , "|", FALSE() )
RETURN
    _toText

     

    Then you can define the following:

    Slope1 = PATHITEM ( [LINEST Output], 1 )
    ResidualSumOfSquares = PATHITEM ( [LINEST Output], 10 )
smpa01's avatar
smpa01
Community Champion
3 years ago

v-jianboli-msft  just being curious here. So if I need Total Sum of Square, Residual Sum of Square and R Square, I need to create 3 seperate measures (TSS, RSS and RSquare) and then create subsequent measures for each  ( incorect TSS, RSS, RSquare) immediately created measures like you did above to derive at the correct TSS, RSS, RSquare value for the filter context?

 

jeffrey_wang 

jeffrey_wang's avatar
jeffrey_wang
Power BI Team
3 years ago

You have to create different measures, one for each statistic you need. There is currently no way to create a DAX function that returns a table value to be shared by multiple measures.

  • AlexisOlson's avatar
    AlexisOlson
    Super User
    3 years ago

    smpa01 You could create a measure that returns a text table using TOCSV and parse that.

     

    For example:

    LINEST Output = 
VAR _filteredTable_ = FILTER ( ALL ( 'fact' ), 'fact'[Country] = MAX ( 'fact'[Country] ) )
VAR _linEst_ = LINESTX ( _filteredTable_, 'fact'[y], 'fact'[x] )
VAR _toText = TOCSV ( _linEst_, , "|", FALSE() )
RETURN
    _toText

     

    Then you can define the following:

    Slope1 = PATHITEM ( [LINEST Output], 1 )
    ResidualSumOfSquares = PATHITEM ( [LINEST Output], 10 )
    • smpa01's avatar
      smpa01
      Community Champion
      3 years ago

      Thanks AlexisOlson , applied here 

      If a linear regreesion is required but only for viz purpose there are tons (I literally mean tons) of javascript library that can do the job at the front-end while developing cust viz and there won't be any DAX dependency.

       

      However, if the linear regression is required as part of the SSAS dataset then DAX is the only option and this is a really nice trick here.