Searching for Patterns
Thomas Chesney · Science Progress · 2011
This book is about how -despite an individual's uniquenesspatterns in behaviour captured inside databases allow organisations to 'group' people.Doing this is usually called data mining.Data mining allows a supermarket to know how many people live in each of its customers' houses (household groups), and how old they are (age groups).It is how the credit card company detect fraud (legitimate transaction group and fraudulent transaction group) and how the bank decides whether or not to give someone a loan (good creditworthiness group and bad creditworthiness group).This book explains how some of the more popular techniques behind data mining work.It does not shy away from the mathematics involved, but does keep all the maths separate from the text in diagrams and tables.The sub-title of the book is 'How we can know without asking'.What this refers to is the power in data mining to generalise from people stored in the database to othersnew customers an organisation has no previous knowledge of.An example used throughout the book is the hugely successful Tesco Club card.The Club card data obviously tells Tesco a lot about its customers, but what many people don't realise is that is also lets Tesco make predictions about new customers who don't own Club cards.This is done by comparing what Tesco do know about a new customer (the contents of one shopping basket for example) with their shopping basket information from their Clubcard data.As a popular science book, Searching For Patters is aimed at the ill defined and difficult to reach, 'general reader'.There are numerous books about data mining, but these generally fall into two categories: text book, highly dense, highly detailed, for serious study of the subject; and how-to guides, such as the Dummies Guide series, which show people how they might implement data mining for maximum business benefit.Searching for Patterns is different.It explains data mining techniques in detail but using non-technical language, and brings in a human interest angle, about the lives and times of the people who laid the foundations that allow data mining to happen.Such people include Claude Shannon, Carl Gauss, Joseph Fourier, Thomas Bayes, Warren McCulloch and Pythagoras, a mixture of famous historical figures and more obscure mathematicians.Searching For Patterns is unlike other popular science books in the number of diagrams, tables and equations which similar books tend to avoid in line with Stephen Hawking's assertion that every equation will halve the readership.