What would be the best approach to analyse accounting data for the Newcomb-Benford’s Law?

It will of course vary by process area and use case, but here is an example of something you could try for O2C data models:

1. Create an OLAP Table and use the LEFT() function to create your dimension as the leading digit of the sales document line item dollar amount (the table/field I usually use is VBAP.NETWR_CONVERTED). PQL Code: LEFT(VBAP.NETWR_CONVERTED || '', 1). Note: Notice that I concatenated an empty string to the dollar amount field by adding "|| ''" to the end. This is required because the LEFT() function only works on Strings and there is no explicit conversion for a float to a string in Celonis PQL. Therefore, we have to do an Implicit Conversion by appending an blank string on the end.
2. In the KPI section of your OLAP table, simply do a COUNT_TABLE(VBAP) to get the count of the occurrence of each number.
3. Sort the COUNT_TABLE(VBAP) in descending order and the numbers should flow from 1-9 accordingly.
4. OPTIONAL: You could now convert your OLAP to a bar chart to more clearly see the distribution of numbers in columnar format.

Hope this helps.

Interesting use case / question, also perhaps you might want to look using Celonis ML https://youtu.be/FJoE2Dbi7Dw

Keep us posted with your finding!