— In the News —
June Huh thought he had no talent for math until a chance meeting with a legendary mind. A decade later, his unorthodox approach to mathematical thinking has led to major breakthroughs.
The current wave of AI startups is made up of businesses that are developing applied AI solutions and they're attracting increasingly larger investments. In this article, Louis Coppey from Point Nine Capital examines 70 applied AI companies that have each raised more than $7 million and creates a framework for success.
— Sponsored Link —
Despite popular belief, the day-to-day work for data science is not always so glamorous. You were promised the role of explorer and storyteller, not report-runner. Regain your path to productivity by enabling your team to find insight on their own. Read why Docusign’s data scientists gave the keys to their product analytics to dozens of teams in the company.
— Tools and Techniques —
The team at Google Design developed these seven key steps for staying user-focused when designing ML-driven products. Rule #1: "Don’t expect Machine learning to figure out what problems to solve." That may sound obvious but there are a lot of key insights throughout this article.
It's common to acquire technical debt when building software and generally, that's okay as long as it's not ignored. But with machine learning systems, technical debt can accumulate very quickly and can kill a fast-moving project. Here's why that happens and what to watch out for on your own projects.
There are many applications for reducing the dimensionality of large datasets. In this post, Elior Cohen explores three commonly used dimensionality reduction techniques: PCA, t-SNE and Auto Encoders. For each, Elior describes how they work and considerations for when to use them. This is easy to follow and includes lots of code snippets and diagrams.
Nice tutorial by Brandon Rose that offers a strategy for making sense of massive amounts of unstructured text. This is easy to follow and includes code.
— Deep Learning —
Insightful post by Pablo Cordero that considers preconceptions about deep learning and the use-cases when deep learning just doesn't make sense.
— Data Viz —
Joyplots are partially overlapping line plots that create the impression of a mountain range. They can be useful for visualizing changes in distributions over time or space. With this new R package, you can make joyplots with ggplot2.
As part of its new People and AI Research Initiative (PAIR), Google open-sourced two new visualization tools to help engineers understand and analyze machine learning datasets. This page offers a project description, live demos and the ability to upload your own datasets.
— Career —
Data science and product management are two of the hottest fields in tech right now. Sometimes the roles overlap but for people with a strong focus on data science, there's a new kind of product manager on the horizon: The Data Product Manager. This post by Trey Causey explores this emerging role and what it will take to succeed.
— About —
Data Elixir is curated and maintained by @lonriesberg. If some awesome person forwarded this issue to you, subscribe for free at dataelixir.com and get it delivered every week.