— Insight —
A lot of questionable numbers are being tossed around lately in the press, social media, and even in the data science community. Sharing data is a natural response in a data-driven world but, especially when the consequences of miscommunication are so high, good data hygiene is crucial. Here's what that means in the context of COVID-19.
Great read about the realities of AI development and the death of an AV startup. There are a lot of issues mentioned here, including the assertion that "supervised machine learning doesn’t live up to the hype."
— Tools and Techniques —
If you have data files that are too large to load into memory, you can use chunking to load subsets of the data into Pandas. But if you want to work with multiple subsets of the data at different times, a searchable index is faster. Here's an easy way to do that with SQLite.
Rob J Hyndman is a statistics professor and co-author of the text, Forecasting: principles and practice. In this post, he explains common forecasting scenarios and why pandemics are particularly challenging.
Great introduction to the field of music modeling in the waveform domain. There's a lot here but it's pretty high-level and organized in a way that makes it easy to skip around.
This webinar is a MUST for companies considering Python for data and analytics projects. Invite your team to learn why Python is popular in the data science community and how companies are reacting to the shift. We'll end with a live demonstration and Q&A. March 26, 12:00 pm ET.
— Resources —
Great thread of book recommendations for sports analytics fans. If you're looking for a good read while you're social distancing or locked down, start here!
— Data Viz —
The Financial Times has been maintaining some of the best charts anywhere on the growth of the COVID-19 pandemic. It's the same data that's being used most everywhere else but as Robert Kosara shows in this post, there are important details that make these charts so useful.
The new, official introduction to D3 is part tutorial and part reference. It's well-organized and built on the Observable platform, which gives you the ability to start with an example and quickly iterate within a live notebook environment.