Book Review: The Signal and the Noise

This review is for Nate Silver’s, The Signal and the Noise : Why Most Predictions Fail – but Some Don’t.

Rating

4.5/5 Very good

I really enjoyed this book.  I felt he had some great example of where prediction works well and where it fails (fitting the subtitle).  I thought the research was excellent.  I definitely walked away learning a lot about statistics and with a few example that I can apply to my own work.

The main facets I didn’t like were his over use of adding in Bayesian statistics and some of the examples were talked through too much.  It’s not that I don’t like Bayesian stats, it’s that I felt like he was adding in weird plugs for it that didn’t always fit in the book nicely.  Had there been smoother transitions, this might have worked.  For the longer examples, I did get tired of reading the same subject after a while.  (That is pretty typical of me though, so that might not be saying much.)

All-in-all, I highly recommend it.  I thought his comparisons, for example: predicting weather and earthquakes, did a great job of demonstrating what makes models feasible or not and how to improve ones you’re currently building.  His writing style is also very enjoyable, filled with smaller relatable situations and fun facts about Nate, himself, or the history of the book’s subjects.  It’s definitely a book for anyone that’s interested in prediction or that works with statistics.