First-digit analysis can be useful for detecting accounting fraud. When numeric data span several orders of magnitude, the first digits of the data are often not evenly distributed among 1 through 9; instead, 1’s can be expected to appear up to 30% of the time (see Benford’s law on Wikipedia). A uniform distribution in the first digits of accounting data can be a first indication that the data was fabricated, but of course, care should be taken in interpretation.
To take the leading digit of a numeric column:
By default, the new column’s name reflects the calculation, but you can give it a new name of your choosing. (Note that changing the name will not affect the calculation itself.)
See also: