Forum Discussion
Benford's law , Power query
- 6 years ago
Hi
please paste this code into the advanced editor and follow the steps:
let MyTable = Table.FromColumns({List.Transform(List.Random(10000), each _ * 10000) } ), ExtractLast2Digits = Table.AddColumn(MyTable, "Last2Digits", each Text.End(Text.From([Column1]),2)), BenfordTable = Table.AddColumn( #table( {"Last2Digits"}, List.Transform( {0..99}, each {Text.PadStart(Text.From(_), 2, "0")})), "BenfordNumber", each 0.01 as number ), #"Merged Queries" = Table.NestedJoin(BenfordTable, {"Last2Digits"}, ExtractLast2Digits, {"Last2Digits"}, "MyTable", JoinKind.LeftOuter), #"Aggregated MyTable" = Table.AggregateTableColumn(#"Merged Queries", "MyTable", {{"Column1", each List.Count(_) / Table.RowCount(MyTable), "ActualDistribution"}}), #"Inserted Subtraction" = Table.AddColumn(#"Aggregated MyTable", "Deviation", each [ActualDistribution] - [BenfordNumber], type number), #"Inserted Absolute Value" = Table.AddColumn(#"Inserted Subtraction", "Absolute Deviation", each Number.Abs([Deviation]), type number) in #"Inserted Absolute Value"It will create a table with the Benford-distribution values.
If you want to apply it to your data, just replace the code in the first step (MyTable) by a reference to your table and make sure that that column with the values to be analyzed is called "Column1". Then you don't have to adjust anything further in the code.
Interesting point on how Benford’s Law can surface anomalies, especially when paired with Power Query for quick pattern checks. For anyone exploring real-world datasets, working with structured legal data can be a practical use case. Resources can offer consistent numerical records to test distributions and assumptions. Applying these techniques to such data can sharpen both analytical and data-cleaning skills.