As I consider myself a hacker rather than a coder I was delighted to find a title aimed for directly at me as "for Hackers". The topic of machine learning is both generally hot and personally interesting to me as I try to mangle recommendation approaches in the music domain and just finished with all worthwhile heuristics-based methods. I thought that this book could provide me with a good entry and could be a newer, more up-to-date approach than the classic Programming Collective Intelligence by Toby Segaran. Nonetheless I was a bit surprised to realize that the hackers' language should be R instead of Python - so this meant that the book would also serve me as an intro to R. As learning R is also on my long to-do list, I was hoping that I could kill two birds with one stone.
The book has a solid structure, the first quarter is covering basic statistics showcasing the capabilities of R. Typical starter exercises of ML are well versed: Bayesian spam filtering, weighting schemes for ranking, regression, overfitting and optimization is explained and demonstrated with a down-to-earth and hands-on approach. The authors also touch some of the more complex topics: principal components analysis, multidimensional scaling and the k-nearest neighbours algorithm are introduced with real world data. I was really delighted to see a longer chapter dedicated to social graph analysis and I totally give kudos for the inclusion of Gephi, the powerful open source graph visualization and manipulation software that I also enjoy using more and more.
The authors come from a social science background and although this book seems to be avoiding 'programming' or providing code that could be incorporated in a runtime environment, I believe the goal they set, to provide machine learning tools for tinkering, has been reached with great success. It's short, concise and not scary at all. Check for yourself at O'Reilly.
Score: 5 of 5.