Hilary Mason on Big Data and Time-Series Analysis at Droptalks
Dropbox runs a monthly tech talk series called Droptalks, which has previously featured speakers like Steve Souders, Guido van Rossum, Greg Papadopoulos, and Amit Singh. A couple of weeks ago, the company hosted Hilary Mason, Chief Scientist at bit.ly, the URL shortener.
Although bit.ly appears to be a simple service, Mason discussed the engineering challenges that arise at such a large scale, as well as the wealth of data the platform generates. Her talk covered the history of bit.ly, the philosophy behind analyzing time-series data, and several engineering tricks, along with demonstrations of three internal tools that will be released as products in the coming months.
Mason's presentation slides are available on SlideShare, and she recommended several introductory books for those interested in analytics and data science:
- Machine Learning for Hackers by Drew Conway and John Myles White, which "uses R on web data and email data."
- Programming Collective Intelligence by Toby Segaran, which she said is "getting a little out-dated, … but it's a really good introduction to how to think about a machine learning program."
- Christopher M. Bishop's Pattern Recognition and Machine Learning, which she described as "the core canonical math book" for those who "just want the math side."
