What Every Data Scientist Should Know About Floating-Point Arithmetic by Michael via @DIYBigData

This is an overview of how the limitations of floating point numbers can (and will) impact data science and other big data applications.

This is essential reading and something I've had to be careful of myself.