— In the News —
A first exploration of a new planet will never happen again in our lifetimes.
The New Horizons spacecraft reached Pluto this week, which is an incredible achievement, in itself, but it's the data that makes it worthwhile. The links below are highly recommended!
More and more decisions, from the music business to the sports field, are being delegated to data. This is a thought-provoking exploration of how well that actually works.
Deep learning is all the rage these days but it wasn't always that way. In the late 90s, it was seen as the stuff of science fiction and was considered to be a dead-end for smart researchers.
This is the story of three computer scientists who chose that path anyway and laid the foundation for the next frontier in AI technology: Geoffrey Hinton, Yann LeCun and Yoshua Bengio. Great read.
— Sponsored Link —
The New Relic Platform lets you collect and display massive amounts of real-time data in a single framework. Every team in your organization can make critical decisions about your software, your customers, and your business—fast.
Create a free account with New Relic and get this swanky shirt for FREE!
— Tools and Techniques —
Great tutorial that teaches backpropagation via a simple python example. This is very well explained and includes worthwhile suggestions for further learning.
Quick review of key linear algebra concepts with examples using Python's numpy package.
— Resources —
Draft version of a free online book that that includes many of the core concepts behind neural networks and deep learning. This "book" uses a hands-on approach and includes a repository of code samples.
Nice collection of R packages for accessing sports data. There's a wide range of sports represented here, including baseball, basketball, the Olympics, cricket, football ("soccer" in the U.S.), American football, hockey, skiing, swimming, etc, etc.
Great collection of organizations, competitions, meetings, and jobs for "doing good" with data science. Many of these would also be great opportunities as career stepping stones.
— Data Viz —
Vega is a visualization grammar - a declarative format for creating, saving, and sharing interactive visualization designs. Vega 2.0 was released this week and includes support for interactivity and streaming data. There's a lot of thought going into this project, it's very well documented and is worth being familiar with.
matplotlib will soon have a new default colormap which is "perceptually-accurate." This short talk describes what that means and why it's important. The changes are motivated by concerns that go beyond matplotlib and should be understood by everyone doing data visualization work. The proposed new colormap is described here >>
Yay! A legend component for D3! If you work with D3, this will be super useful.