5 of 5 people found the following review helpful
When it comes to the crunch,
Verified Purchase(What is this?)
This review is from: Super Crunchers: How Anything Can Be Predicted (Paperback)
I came across a reference to this book in a news article about Epagogix, a company that analyses some of the characteristics of film scripts in an attempt to identify whether a film will be successful or not. The author uses Epagogix as an example of the way in which simple statistical models are increasingly being applied to large data sets in order to improve decision making in a range of business and social applications. The interesting point is that the models often outperform the judgement of human experts in tasks such as pricing airline seats, diagnosing and treating disease, finding a life partner and paroling prisoners.
One of the other example tasks he cites is the choice of book titles, and demonstrates it with an account of how he used the statistical technique of randomized testing with Google AdWords to choose between "The End of Intuition" and "Super Crunchers", based on the number of clicks each received. This - perhaps inadvertently - isn't really a good advert for the technique, since the former title actually means something, while the one he ended up using doesn't (what's the difference between a super cruncher and a regular, standard, non-super cruncher anyway?).
However, its use is in keeping with the author's breezy, non-technical style (e.g. "He's the kind of guy that would quickly disabuse you of the notion that number crunchers are meek, retiring souls. I've seen [him] stride around a classroom, gutting the reasoning behind a seminar paper with affable exuberance" [p2]) that tries to spice up what could have been a dull subject in other hands. Thus, he describes the techniques - regression analysis and randomized testing - briskly without any mathematics (there are only a couple of diagrams in the entire book) before moving onto stories about their application. Most of the technical detail and justification has been moved to a notes section that takes up more than a tenth of the book.
I enjoyed reading this account, particularly his exploration of why this sort of thing is becoming more widespread (increasing computer power, cheaper storage, wider accessibility via the internet) and its implications for privacy and control. Those who found Freakonomics stimulating will, I think, like this as well (it should be noted that the present author has worked with one of that book's coauthors - in the past), since it provides some intriguing insights into the way in which trends in the world and our behaviour in it can be unraveled and predicted.