Review: Python for Data Analysis
I’ve been using Python for quantitaive economics and geography for the last five years and I wish this book had been written back then. It is clear, easy to read, full of (cool) examples and incredibly useful, regardless of whether you read it from first to last page or have it around to check while hacking. If you are getting started on data crunching with Python, you should not miss it. If you are not new to the party, this will still be a good resource to get your head around a bit more advanced pandas, and the chapter on advanced Numpy is very informative too.
However, if you are over most of the learning curve, the book has a different feel. Although the title might suggest different, “Python for Data Analysis” is mostly “Pandas for Data Analysis”. Fair enough, if you’ve written the library itself, it seems appropriate to focus on it, but I think it should be stressed from the beginning. Other crucial elements of the Python data stack (matplotlib, statsmodels, sklearn…) are sometimes touched upon, but only briefly, and only to the extent you can relate them to pandas. I would have enjoyed it much more if those were covered a bit more in detail. Still, very good book.