Rank over partition multiple columns in Power Query NOT DAX!

Hi patoduck ,

yes, please try the following code (paste into the advanced editor and check what it does):

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WMlTSAWMDPSMLSwMzI1OlWB2IqBFE1NTcwhBJ1BgiamRmiqzWBEPUCGGusamFiaUhXBRmrom5oTlc0BirUqix5iBhpdhYAA==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type text) meta [Serialized.Text = true]) in type table [document = _t, topic = _t, gamma = _t]),
    #"Changed Type" = Table.TransformColumnTypes(Source,{{"document", Int64.Type}, {"topic", Int64.Type}, {"gamma", type number}}),
    #"Grouped Rows" = Table.Group(#"Changed Type", {"document"}, {{"All", each Table.SelectRows(_, (x) => (x[gamma] = List.Max(_[gamma]))){0}}}),
    #"Expanded All" = Table.ExpandRecordColumn(#"Grouped Rows", "All", {"topic", "gamma"}, {"topic", "gamma"})
in
    #"Expanded All"

If your raw data has all rows for one document in a sequence, you can even use GroupKind.Local to speed up the calculation considerably:

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WMlTSAWMDPSMLSwMzI1OlWB2IqBFE1NTcwhBJ1BgiamRmiqzWBEPUCGGusamFiaUhXBRmrom5oTlc0BirUqix5iBhpdhYAA==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type text) meta [Serialized.Text = true]) in type table [document = _t, topic = _t, gamma = _t]),
    #"Changed Type" = Table.TransformColumnTypes(Source,{{"document", Int64.Type}, {"topic", Int64.Type}, {"gamma", type number}}),
    #"Grouped Rows" = Table.Group(#"Changed Type", {"document"}, {{"All", each Table.SelectRows(_, (x) => (x[gamma] = List.Max(_[gamma]))){0}}}, GroupKind.Local),
    #"Expanded All" = Table.ExpandRecordColumn(#"Grouped Rows", "All", {"topic", "gamma"}, {"topic", "gamma"})
in
    #"Expanded All"

But this only works if they are all nicely in one sequence each.