Although implemented in the Test class, the Kolmogorov-Smirnov statistic is known to work for continuous distributions, not discrete ones, such as the ones from the digits tests (1, 2, 3…9, and so on).
The Mantissas test, however, is made on a continuous distribution, since the ordered mantissas of the log10 of a Benford set numbers are expected to form a straight line from 0 to almost 1. So far, the mantissas test is more of a qualitative one, from eyeballing the plot. We should implement and test KS for the mantissas, and check it’s prospects.
Although implemented in the Test class, the Kolmogorov-Smirnov statistic is known to work for continuous distributions, not discrete ones, such as the ones from the digits tests (1, 2, 3…9, and so on). The Mantissas test, however, is made on a continuous distribution, since the ordered mantissas of the log10 of a Benford set numbers are expected to form a straight line from 0 to almost 1. So far, the mantissas test is more of a qualitative one, from eyeballing the plot. We should implement and test KS for the mantissas, and check it’s prospects.