Issue 157
— In the News —
How Facebook’s Oracular Algorithm Determines the Fates of Start-Ups
Facebook is remarkably effective at turning algorithms into dollars. The platform is so good at “microtargeting,” many small companies don't even bother to advertise anywhere else. But for many advertisers, there are downsides to all the levers that Facebook's vast machinery offers. This article in the New York Times explores the human side of collaborating with an AI.
SignalBox is a preconfigured Deep Learning Platform that lowers the cost of experimentation and reduces your need for engineering support. This is a super useful platform that I've become personally involved with. SignalBox supports a wide variety of clients in AdTech, FinTech, Energy, and Customer Service. Check out some of the current use cases >> Upcoming features include robust image diffing at scale and OCR to support industrial applications and geospatial analytics. For more info, send a note to [email protected] or set up a time on my calendar and let's chat.
— Sponsored Link —
SignalBox - Deep Learning Made Simple
If you ever need 3D plots in Jupyter, here's a great option to consider. This notebook demonstrates how to integrate MathBox plots into Jupyter while getting data via Python. The examples in this notebook are interactive and presentation-quality. Last week, Twitter increased the character limit on tweets from 140 to 280. For users, it's a big deal. An analysis of the decision was posted on Twitter's engineering blog, which led Jake VanderPlas to question if a similar analysis would apply to Python line length limits. Don't be put off by the title. This post by Jake is a fantastic exploration of the question. Simon Willison, co-creator of Django, just released Datasette, an open-source tool that will easily create and publish an API for your SQLite databases. Datasette enables you to package your datasets into a simple, well-understood format and then deploy them in a read-only, highly scalable way. Combined with Simon's csvs-to-sqlite, you can quickly turn a CSV file into a live-on-the-internet REST API that supports arbitrary SQL queries. This will be useful. Adam Kelleher from Buzzfeed has been writing a series of posts about causality. In this latest post, Adam describes the thinking behind a new Python package that will help make causal inference easy for data analysts and scientists. Here's a curated collection of worthwhile Python frameworks, libraries, software and resources. This is thorough and very well organized. For a curated collection of R packages and tools, see Awesome R >>
— Tools and Techniques —
Jupyter meets MathBox2
Exploring Line Lengths in Python Packages
Datasette
Causal Inference With pandas.DataFrames
🕶️ Awesome Python
Neural networks are not just another classifier, they represent the beginning of a fundamental shift in how we write software. They are Software 2.0. Read this article by Andrej Karpathy first, then read Pete Warden's take on the same idea: "Deep Learning is Eating Software." Beautiful article in Distill that shows how neural nets build up their understanding of images.
— Deep Learning —
Software 2.0
Feature Visualization
Coursera has a new online course starting soon that's worth checking out: How to Win a Data Science Competition: Learn from Top Kagglers. Kaggle is well-known for its data science competitions and as this recent article in Wired attests, competing on Kaggle can help your job search. Kaggle has a lot going on and is worth spending some time with. If you're interested in a quick start guide to the competitions, check out DataQuest's Kaggle Fundamentals tutorial. Looking for an AI or machine learning role at a startup? Daniel Gross, a partner at YCombinator, suggests some key questions to ask your interviewers about their business.
— Career —
Kaggle Guides
Business questions you should ask when interviewing at ML/AI companies
Be sure to catch the most popular links from last week's issue...
— In Case You Missed It —
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.
— About —
Sign up for Free
and join the thousands of data lovers who already start their week with us.
No spam, ever. We'll never share your email address and you can opt out at any time.
No spam, ever.