1 of 133

Mathematics Seminar

Mahidol University International College, Salaya, Thailand

Benford’s Law and GLORQ

Wed, September 20, 2023

2 of 133

Lecturers:

(1) Alex Ely Kossovsky

(2) Wayne Lawton

3 of 133

=============================

4 of 133

Simon Newcomb

5 of 133

Simon Newcomb - Canadian-American astronomer, physicist, and mathematician.

Note on the Frequency of Use of the Different Digits in Natural Numbers

Published in 1881.

6 of 133

Frank Benford

7 of 133

Frank Benford – an American physicist and electrical engineer.

The Law of Anomalous Numbers

Published in 1938.

8 of 133

9 of 133

The first digit on the left side

10 of 133

Mother duck leads her kids.

She is the first.

11 of 133

Natural Leader!

12 of 133

The first digit on the left most side

13 of 133

Data Set

14 of 133

Focus on 1st digits

15 of 133

16 of 133

What proportions should we expect?

17 of 133

Data is random by definition.

So the first digit should be random too!

Should we then have equal probability for all digits?

18 of 133

NO!

Digits do not occur equally!

19 of 133

=============================

20 of 133

Let’s make an empirical real-life data experiment on digits:

21 of 133

Data on Molar Mass

The molar mass values of 32 randomly chosen chemical compounds/molecules:

22 of 133

23 of 133

24 of 133

25 of 133

26 of 133

Results:

Digit: { 1, 2, 3, 4, 5, 6, 7, 8, 9 }

Count: { 8, 6, 4, 5, 2, 3, 2, 1, 1 }

27 of 133

Digits are NOT equally distributed!

Low digits such as {1, 2, 3} are frequent.

High digits such as {7, 8, 9} are rare.

28 of 133

=============================

29 of 133

Benford’s Law

Probability of digit d = LOG10(1 + 1/d)

d = { 1, 2, 3, 4, 5, 6, 7, 8, 9 }

30 of 133

Probability of digit d = LOG10(1 + 1/d)

LOG10(1 + 1/1) = Log(2.00) = 0.301

LOG10(1 + 1/2) = Log(1.50) = 0.176

LOG10(1 + 1/3) = Log(1.33) = 0.125

LOG10(1 + 1/4) = Log(1.25) = 0.097

LOG10(1 + 1/5) = Log(1.20) = 0.079

LOG10(1 + 1/6) = Log(1.17) = 0.067

LOG10(1 + 1/7) = Log(1.14) = 0.058

LOG10(1 + 1/8) = Log(1.13) = 0.051

LOG10(1 + 1/9) = Log(1.11) = 0.046

---------

1.000

31 of 133

Benford’s Law

1st Digits

32 of 133

33 of 133

Data is random,

but 1st digit is not so random!

34 of 133

Let’s empirically examine some real-life data sets.

35 of 133

36 of 133

37 of 133

Benford’s Law in Finance

Empirical evidence from price catalog data for big supermarkets, shops, and corporations.

Data: The entire catalog (prices in $) for all the products on sale by Canford Audio PLC in the U.K. which manufactures and retails 15194 electronic items.

http://www.canford.co.uk/

38 of 133

Canford Audio PLC - UK

39 of 133

Catalog

40 of 133

41 of 133

42 of 133

Canford Audio PLC [Entire Data]

43 of 133

Canford Audio PLC [Entire Data]

44 of 133

1st digits - Canford Catalog

45 of 133

=============================

46 of 133

All 3338 stocks traded on the New York Stock Exchange on 23 December, 2011.

Price and Volume of Stocks are recorded.

47 of 133

48 of 133

49 of 133

Low digits such as { 1, 2, 3 } are frequent.

High digits such as { 7, 8, 9 } are rare.

50 of 133

Volume Traded

51 of 133

Volume Traded

52 of 133

Price of Stocks

53 of 133

Price of Stocks

54 of 133

===============================

55 of 133

NASDAQ USA Market Capitalization

(Honest Data)

Oct 9, 2016

2,889 companies

http://www.nasdaq.com/screening/companies-by-industry.aspx?exchange=NASDAQ

56 of 133

57 of 133

NASDAQ Market Capitalization

58 of 133

NASDAQ Market Capitalization

59 of 133

===============================

60 of 133

Oklahoma State Expenses

(Honest Data)

The State Of Oklahoma in the USA 966,990 payments in 2011.

https://data.ok.gov/dataset/state-oklahoma-vendor-payments-fiscal-year-2011

61 of 133

62 of 133

Oklahoma overall 1st digit configuration:

Nearly Perfect compliance with Benford!

63 of 133

Oklahoma overall 1st digit configuration:

64 of 133

Near perfection !

65 of 133

=============================

66 of 133

Benford’s Law in Geology

Empirical evidence from Earthquake data.

Evidencia empírica de los datos del terremoto.

Data: The measurements of time intervals (in seconds) between all the 19,453 earthquake occurrences worldwide in the year 2012 - regardless of magnitude.

67 of 133

68 of 133

69 of 133

The United States Geological Survey Organization

http://earthquake.usgs.gov/earthquakes/eqarchives/epic/

70 of 133

71 of 133

72 of 133

73 of 133

Earthquake & Benford

74 of 133

Near perfection !

75 of 133

========================

76 of 133

Benford’s Law in Astronomy

77 of 133

78 of 133

Stars within our home, the Milky Way Galaxy.

79 of 133

Random sample of 50,000 stars in the Milky Way Galaxy.

Variable to measure: distance from the Solar System in Light Years.

80 of 133

Source: NASA

http://heasarc.gsfc.nasa.gov/W3Browse/star-catalog/hipparcos.html

81 of 133

{28.3, 15.1, 12.0, 10.5, 9.0, 7.6, 6.5, 5.9, 5.2}

82 of 133

Star Distance & Benford

83 of 133

=============================

84 of 133

Benford’s Law in Geology

85 of 133

Earth Anisotropy & Benford

86 of 133

Geomag Field & Benford

87 of 133

Near perfection !

88 of 133

CONCLUSION:

Low digits such as { 1, 2, 3 } are frequent.

High digits such as { 7, 8, 9 } are rare.

89 of 133

====================

90 of 133

Benford’s Law as a

Tool for Fraud Detection

91 of 133

Cheating – Fake Data

Digital Equality

92 of 133

Honest – Real Data

Digital Inequality (apprx like Benford)

93 of 133

Let us examine revenue data from 5 different companies.

4 companies are honest, reporting real and authentic revenue data.

1 company is dishonest and cheating, reporting fake data so that they could pay less tax.

Can you discover the cheating company?

94 of 133

95 of 133

Have you decided who is the guilty one?

Have you discovered fraud?

--------------------------------------

Let’s focus on 1st digits:

96 of 133

97 of 133

The cheating company should be obvious now!

98 of 133

First Digits Proportions

99 of 133

It seems that Nokia has committed fraud!

100 of 133

====================

101 of 133

Election Fraud Detection

102 of 133

103 of 133

but….

104 of 133

There is a pitfall:

105 of 133

The well-educated, smart, and corrupt cheater, already knows about Benford!

106 of 133

We can detect such sophisticated fraud via a patented algorithm.

107 of 133

Data Forensics via:

Digital Development Pattern

108 of 133

Benford’s Law is a statement about the digital configuration of the entire data set.

109 of 133

Could we then naively conclude that any sub-set of the data on any segment of the x-axis is also Benford?

110 of 133

Does the whole endow its Benford property to its parts?

-----------------------------------------------------

NO!

111 of 133

Local digit configuration changes and develops as we move from the left to the right on the x-axis!

112 of 133

Cut your data into several pieces along the x-axis with an imaginary scissor:

113 of 133

Magical Mathematical Scissors!

114 of 133

115 of 133

Digital Behavior

Political Narratives

=

By some amazing coincidence…

116 of 133

117 of 133

….but….. where to cut?

118 of 133

Only along powers of 10!

119 of 133

This can ONLY be seen on sub-intervals between

… 0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000, 100000, …

120 of 133

This phenomenon is called:

Digital Development Pattern

121 of 133

All honest data should come with Digital Development Pattern!

122 of 133

Patent # 9,058,285 has been granted for this algorithm by the US Patent Office in 2015.

123 of 133

124 of 133

125 of 133

Patent’s Title:

Method and system for Forensic Data Analysis in fraud detection employing a digital pattern more prevalent than Benford’s Law”.

126 of 133

====================

127 of 133

Serial numbers

code numbers

index numbers

Passport number

ID number

Driver License number

Artificial numbers

ATM withdrawal amounts

All are not Benford, and here digital equality is the norm.

128 of 133

Scale Invariance Principle

The law is (almost) immune to changes in scale!

The law is (almost) immune to changes in units!

129 of 133

Digits of data on time between 19,453 earthquake occurrences in 2012 are almost the same for all time scales.

130 of 133

Canford Catalog in Euros:

1 Euro = $1.1035 Dollars

131 of 133

132 of 133

This flexibility endows Benford’s Law a measure of universality!

133 of 133

=============================