Kaiser Fung

Libro muy, muy interesante en el que en cuatro grandes bloques el autor analiza los usos cotidianos de la estadística, más allá de lo que solemos saber. La estadística se usa cuando aparece un brote epidemiológico, cuando diseñas análisis de dopaje, o exámenes de polígrafo, cuando intentas minimizar los tiempos percibidos por los visitantes de Disneylandia o los viajeros de una red de autopistas, o cuando intentas diseñar un examen tipo test que no discrimine a los estudiantes según su origen. El autor nos muestra bastantes veces que lo que dicta nuestro sentido común es un craso error. He tenido que releer varias veces varios pasajes porque lo que leía me parecía claramente incorrecto, pero resulta que se ha comprobado que hay cosas que son como son a pesar de nuestros prejuicios. El autor se esfuerza en hacernos entender que, por ejemplo, dada una fiabilidad de un test (por ejemplo, el test de dopaje por nandrolona es fiable al 99,9%), cualquier esfuerzo que hagamos por evitar falsos negativos (es decir, por impedir que se escapen los dopados) incrementará el número de falsos positivos (gente limpia que da positivo en el test) y que según el contexto nos interesará más mejorar uno de los dos números (un falso negativo hace trampas pero no sale en los periódicos, un falso positivo es una serie de juicios y portadas en los periódicos criticando el sistema y proclamando su inocencia).
El libro es muy interesante, reitero, y va un paso más allá de la introducción a los temas, siendo en ocasiones lo suficientemente denso como para tener que dedicarle ratillos a digerir las ideas del autor. Acaba el libro con una gigantesca bibliografía comentada que da para cientos de horas de lectura adicional.
Me ha encantado.… (mehr)

“The key is not how much data is analyzed, but how.”

Data is manipulatable. The same set of data can be analyzed to give the exact polar results. With the accessibility of the Internet, we are living in a world of lots of data. “Big Data” is the word the author used. It’s a vast number of data that’s beyond the scope of any normal data analysis program can handle or manage. Lots of data are obtainable, with lots of analyses of these data available, since every single one of the market players are studying these data to gain an edge in competition.

The author used the Gates Foundation’s example to let us know that even big organizations with lots of money and analysts can still make a stupid decision with the wrong data or analysis. Ten years ago, the Foundation made a mistake assuming that smaller school s are better for student achievement, which is later proven untrue. He argued that Big Data moves us backwards, since more data results in more time spend analyzing, arguing, validating and replicating results. More of the any above activities will cause more doubt and confusion. Therefore, It’s urgent to learn a way to analyze them so you can just keep your head clear, and not being lied to.

“Any kind of subjective ranking does not need to be correct, it just has to be believed”

What do we believe, and what technique do we use to help us make the decision? Data analysis is an art, and not every statistician knows what he’s talking about. A person with good “numbersense” will be way above the others in avoiding the pitfall. A person with a good numbersense will spot bad data or bad analyses, or know when to stop when collecting his own. Unfortunately, numbersense can’t be taught in a regular classroom, a program or a textbook. It’s only learned from another person or real life practices. After more than 20 years in management in a hospital, I know these people do exist, but rarely. They are wonderful problem solvers. Lucky for rest of us, this book is a great place to start learning about numbersense. The author has a way of explaining complex subjects in a simple and understandable way, and his flow of thoughts is logical and very easy to follow. While analyzing data, the author also explained statistical terms thoroughly, as the term significant does not necessarily means important.

The author used real life news examples where someone made a claim about something and then backed it up with data, and he analyzes them, explaining the process to us along the way. The examples include: Law schools admission data, Groupon’s business model, diets and BMI, unemployment and jobs, our inability to remember prices and CPI, and even fantasy football. These examples were very interesting to read as the author gives step-by-step instructions of how these data we see everyday could easily be manipulated to fool us. My daughter is in the process of applying for college, and I can assure you, after reading the first chapter, I will never look at college rankings the same way.

I think every person in marketing, business, sociology, management or data analysis should read this book, as well as any consumer who wants to make sense of this so called “Big Data.” Numbersense is a great word for people who have the talent of analyzing data and spotting errors or intended manipulation. This book reads very much like [b:Freakonomics: A Rogue Economist Explores the Hidden Side of Everything|1202|Freakonomics A Rogue Economist Explores the Hidden Side of Everything|Steven D. Levitt|http://d202m5krfqbpi5.cloudfront.net/books/1327909092s/1202.jpg|5397] by Stephen Levitt, but is a bit more technical and might take a little understanding of statistics and/or business to fully appreciate the book. My background is business administration and healthcare, and I had a fun ride.

*Thanks to Netgalley and McGraw Hill in providing the advanced reading copy.… (mehr)

Have you ever read a news report of a new study or statistic and felt instantly skeptical of the findings? These days, information is everywhere, but if you don’t know how to interpret it or at least read it properly, it can become twisted to support many different theories. Kaiser Fung’s Numbersense is an attempt to teach readers just how to mine large sets of data for relevant, true, and reality-based conclusions. While it may not be completely relevant to your life, it does offer a new way of looking at the world.

Fung takes examples from current events and stories to show how data is presented to the public and just how it is derived and manipulated. He looks at the following models:

• Law schools, and statistical manipulations used to increase their national rankings
• BMI calculations, and how differing measures can lead to different health findings
• The Groupon phenomenon, and how it actually hurts local businesses
• Internet marketing initiatives, and how false positives lead to more spam
• Unemployment rates, and how seasonality can skew the public’s perception of the economy
• The Consumer Price Index, and how averaging disparate entities can cause miscalculations
• Fantasy Football Leagues, and how balance beats flash on the fantasy field

In each of these examples, Fung delves deep into the data to find interesting areas where the common perception can be skewed by how the data is analyzed. The weird thing I kept thinking was if Fung thinks that most data presented is skewed or flawed in some way, how are we to trust him? Isn’t he also presenting seemingly authoritative data? Throughout the book, he touts the quality of “numbersense” (constantly presented in small caps in the text). It’s almost as if he’s trying to sell a new weight loss system or tax program. In the end, though, his examples do lead to new ways of looking at data. This is indeed the era of Big Data; learning how to understand it not a bad skill to have. This book will definitely be of interest to analysts and skeptics, but anyone looking to peek behind the statistical curtain will get something out of it. A curious and quick read.… (mehr)