— In the News —
A data scientist at Twitter reflects on what a "data scientist" really does. There are a lot of great insights here.
Economics evolved without access to great data and the result has been a different set of values and conventions than that used in other number-intensive disciplines. That's changing fast and Economics is starting to look like other fields that rely on data.
This article explores what users really want from their data and how to give it to them. It's not charts and fancy graphics.
— Sponsored Link —
This eBook shows you how to build great push messages that convert for your app.
— Tools and Techniques —
“A Neural Algorithm of Artistic Style” was recently posted to ArXiv, featuring some of the most compelling imagery generated by deep convolutional neural networks since Google Research’s “DeepDream” post. This article by Kyle McDonald examines the initial public implementations of this paper, with links to the Github repos.
proof is a Python library for creating optimized, repeatable and self-documenting data analysis pipelines. proof was designed to be used with the agate data analysis library, but can be used with numpy, pandas or any other method of processing data.
Series of tutorials that compares Python and R implementations for common data science tasks.
— Data Viz —
Should your charts always start with a zero baseline? This is a great article by Nathan Yau that explores the question with convincing graphics and options.
This article explores a technique for visualizing many events across multiple timescales in a single image, where little or no zooming is required. It's an interesting technique and the article is very well written with sample code and lots of examples.
— Byte Sized —
We need new words to describe the coming wave of machine-generated information.
To track and monitor the population, right whales are photographed during aerial surveys and then manually matched to an online photo-identification catalog. This competition aims to automate that process.