Return to Programming Book Reviews.
Data Mining and Statistical Analysis Using SQL
by Robert Trueblood, John Lovett, Jr.

Publisher: Apress Buy it Now! 

ISBN: 1893115542
Format: Paperback, 500pp
CD: NO

As the cover says, this book isn't designed to be just another theoretical text on statistics and data mining. Instead, it's aimed at database administrators and software developers who want to buttress their understanding of statistics to support data mining and customer relationship management analytics, and who want to use SQL for these tasks. So, you are faced with the tasks like analyzing Web logs or customer behavior patterns. Considering the amount of hype surrounding data mining and related fields, chances are that you are looking for the right book, packed with "how to" information for professionals who may poses minimal statistical knowledge. Depending on specific needs and previous knowledge, this may be just the right book for all "ordinary" DBAs out there who want to become proficient with applied mathematics - statistical analysis to be more precise.

The only problem with this book is closely related with data mining hype mentioned above. The words "data mining" from the title are a bit misleading - it describe basic statistics, but from my perspective it cannot really be qualified as a data mining book. Don't get me wrong, the techniques described inside are very valuable for various DM tasks, especially in data preparation and analysis phases. If you need more than that, and have the necessary experience and skills, go for another, more DM-focused book. But if you need a "no nonsense" book that will guide you through the practical statistical techniques without delving deeply into the theoretical foundations, this one is the best I've seen in years.

Each chapter is independent and self-contained, so there's no need to go through all 400 pages (short SQL and VB references are also included) before you can start applying data analyses. Every technique is introduced by formulas that illustrate the required analysis, few numerical examples from a "real world" database and various visualization options. At the end of each chapter, you'll get a working SQL code (MS Access, easily translated to other SQL flavors) that can be easily used in your solutions. At a glance, the book covers basic statistical principles (including a diagnostic tree), measures of central tendency and dispersion, goodness of fit, additional tests of hypothesis, curve fitting, control charting, analysis of experimental designs, time series analysis, and offers an overview of relational database and SQL principles, various statistical tables and related Visual Basic routines.

I must say that I enjoyed reading this book, and this feeling remained after I tried to apply some of the techniques to real-world problems. aboutAI.net uses Data Mining extensions for SQL Server extensively, and this book really helped me to understand some statistical concepts and solve several practical issues. Authors are both working analysts with distinguished academic backgrounds, and their approach is to explain terms and concepts gradually: starting with prose, presenting the appropriate graphs, continuing with formulas and finally producing understandable SQL queries. The result: you will not feel overwhelmed, but motivated to try it out for yourself.