FREQUENCY() returns the expected Benford frequency for sequential leading positive numeric digits to a precision of eight digits. It lets you perform limited Benford tests for specific situations.
Use FREQUENCY() in place of a full Benford analysis when you only want to focus on specific digit combinations. For example, when auditing insurance claims that have approval limits at specified claim amounts, you could use the FREQUENCY() function to investigate amounts just under an approval threshold. To investigate claims valued close to an approval limit of $5,000, you could select the range from $4,900 through $4,999. First, count the total number of records, then use a filter to count the records for which LEADING() returns 49 and compare the ratio of the two counts to the value you get for FREQUENCY(49). This is faster than running a complete analysis on a data file of a million records and it does not generate a large data file or lengthy entries in the Command Log.
FREQUENCY(C)
| • | C is a character string that consists only of the digits 0 to 9 and has an overall value greater than zero. |
FREQUENCY() ignores any leading zeros in the character string.
Note: Specifying strings longer than six digits can result in zero values. Calculations for strings longer than six digits may require greater precision than Analyzer’s limit of eight decimal places.
FREQUENCY("43") =.00998422
FREQUENCY("87654321") = 0.00000000
The frequency of 87654321 is 0.00000000495. Analyzer computes to a precision of eight decimal places and therefore returns a zero value.