Sorting rows based on two columns

[Fda]

Hi!

I am struggling with a data exercise and could use some help.

I would like to return the value of column A, based on a grouped value of column B, where the value of column C is the highest.

Example:

Column B has the value 600001 in the first three rows, and the values 1, 2, 3 in column C.

The value I would like to sort out from column A is therefore 4223900.

I have tried conditioned formatting, returning max values, most anything shy of a manual sorting, but no luck.

Thanks!

[BekahLoSurdo]

Hi @Fda,

Try this:

let
    Source = Excel.Workbook(File.Contents("C:\Users\rlosurdo\Desktop\GroupByColumn.xlsx"), null, true),
    Groups_Sheet = Source{[Item="Groups",Kind="Sheet"]}[Data],
    #"Promoted Headers" = Table.PromoteHeaders(Groups_Sheet, [PromoteAllScalars=true]),
    #"Changed Type" = Table.TransformColumnTypes(#"Promoted Headers",{{"ColumnA", Int64.Type}, {"ColumnB", Int64.Type}, {"ColumnC", Int64.Type}}),
    #"Added Custom" = Table.AddColumn(#"Changed Type", "MaxGroupValue", each let group = [ColumnB] in
        List.Max(Table.SelectRows(#"Changed Type", each [ColumnB] = group) [ColumnC])),
    #"Filtered Rows" = Table.SelectRows(#"Added Custom", each [MaxGroupValue] = [ColumnC]),
    #"Removed Other Columns" = Table.SelectColumns(#"Filtered Rows",{"ColumnA"})
in
    #"Removed Other Columns"

[Fda]

Thank you for taking your time, and it looks like its working!

Is it a heavy calculation for it to make? It's running pretty slow for me, and I've got a lot of rows left?

[BekahLoSurdo · Accepted solution]

Hi @Fda,

Without seeing your dataset, I'm not sure about the exact performance. If you want to try another way (and compare the two for your particular dataset), this can also be done outside of Power Query by creating a new column with DAX:

ColumnA_IfMax = IF ( CALCULATE ( MAX ( 'Table'[ColumnC] ), ALLEXCEPT ( 'Table', 'Table'[ColumnB] ) ) = 'Table'[ColumnC], 'Table'[ColumnA], BLANK() )

[Fda]

You are a legend, that is exactly what I was looking for!

Thank you very much!

[BekahLoSurdo]

Glad to help!