— In the News —
FiveThirtyEight is making available nearly 3 million tweets that were reportedly created to interfere in the U.S. presidential elections and later deleted. Given recent assertions that social media "trolls" are already working to undermine upcoming elections, this is an important and timely dataset. This feature story provides background information about the data, what it shows, how it was obtained, and how to access it for your own analysis.
Facebook provides a variety of tools to help communities recover from natural disasters. It's a worthy effort but what's really interesting here are the types of data that they're able to derive from user engagement on the app. Facebook-level scale is an amazing thing.
— Sponsored Link —
Udacity’s School of Data Science offers a range of compact and efficient online learning programs that are built with industry leaders like IBM, Kaggle, Tableau, and Mode.
Whether you are just getting started in data, are looking to augment your existing skill set, or intend to pursue advanced studies and career roles, Udacity has the right learning program for you. Take advantage of world-class curriculum and a flexible learning model to advance your career with Udacity.
— Tools and Techniques —
In this case study, Yotam Drechsler from BrainQ walks through their approach for successfully making data science experiments agile.
There's been a lot of research, discussion, and tools that are intended to help data scientists see inside their models but transparency isn't always needed. This is a great post that explores how black your black-box should really be.
For people who didn't spend a lot of time studying math and statistics in school, here's what you need to know to get started in machine learning. Covers what to learn and how to learn, including links for suggested courses, texts, and resources. This post has gotten a lot of attention around the web this week.
L1: Tensor Studio is a live-programming environment for differentiable linear algebra. It's a combination of a programming language, interpreter, standard library and IDE in one package. Think of it as a playground for tensors.
A/B testing may sound easy but there are a lot of things that can go wrong. In her latest post, Emily Robinson offers 12 key guidelines for making sure your tests are successful. Each guideline includes useful rationale, linked references, and screenshots where applicable.
— Data Viz —
Dashboards seem to be everywhere and for a variety of reasons, the authors of this survey paper are calling it a "dashboard conspiracy." This is a worthwhile project that explores how dashboards are used in practice, dashboard design, decision support, noteworthy examples, limitations, and challenges.
This guide to using color in data visualization is well organized and includes an awesome assortment of tools and linked references along the way.