— In the News —
Interested in the business side of AI? Bradford Cross is a founding partner at DCVC (Data Collective), the world's leading machine learning and big data venture capital fund. His insights are informed and worthwhile. In this post, Bradford offers some good directions to explore and some warnings. From his perspective, "2017 will be the year of reckoning."
Deepmind is developing a blockchain-like project that will allow hospitals, and eventually patients, to see exactly who is using health-care records and how. The idea is to create an indelible audit trail that will help build trust and accountability. It's being developed specifically for health data but it's easy to see how this idea could be extended.
Big news this week... Kaggle is joining Google Cloud. If you're not familiar with Kaggle and don't understand what all the fuss is about, check it out: www.kaggle.com. Kaggle is a large community of data scientists and is designed to help you "learn, work, and play."
— Tools and Techniques —
The Stitch Fix Data Science team works on some super interesting problems and this animated scrollytelling article, offers an awesome overview of what they're up to.
Great post by Jake VanderPlas that details his workflow with Jupyter Notebooks. Jake is a prolific contributor to the data science community and is the author of the recently released, Python Data Science Handbook.
The principles behind neural networks are actually fairly simple. Here's a gentle walk-through that shows how to use deep learning to categorize images.
tidyquant bills itself as "a one-stop shop for serious financial analysis." Along with retrieving stock prices, key stats, financial statements, dividends, etc from the web, tidyquant will put that data into a "tidy" dataframe so you can perform complete financial analyses in the tidyverse. This latest release adds performance analytics and improved documentation.
— Resources —
This course at Stanford is in progress now and all slides and lecture notes are online and freely available.
— Data Viz —
geoplot is a high-level Python geospatial plotting library. It's an extension to cartopy and matplotlib which makes mapping easy: like seaborn for geospatial.