Keeping up with data — Week 51 reading list
5 minutes for 5 hours’ worth of reading
The image above is from Bill Schmarzo’s article about data monetisation. A topic that has become very close to my heart this year and is a key focus in my new business venture. Indeed, data has the potential of being a very valuable asset. But it has no value sitting on servers. It needs to be put to use. Let me quote from the article:
Companies must transition their executive mindset from “data as a cost to be minimized” to “data as an asset that will fuel the economic growth of the 21st century.”
With that in mind, let’s get into the best articles I’ve read this week.
- From 0 to 60 (Models) in Two Years: Building Out an Impactful Data Science Function: What were the factors behind the success of WW’s data science team? Executive support, pragmatic approach and project road map, a well-built team, close cooperation with dev team and partnership with product team. I very much identify with these points as I’ve experienced what happens if any of them is missing. Great tips if you are starting a data science team. But equally useful to compare and contrast with existing ones. (Carl Anderson @ Medium)
- Production Machine Learning Monitoring: Outliers, Drift, Explainers & Statistical Performance: Monitoring of ML models in production has been a big topic of 2020 (at least for me). The article presents an end-to-end example showcasing best practices, principles, patterns and techniques of ML monitoring. Read more about performance, statistical, outlier detection, concept drift and explainability monitoring. (Alejandro Saucedo @ Towards Data Science)
- Architecture for High-Throughput Low-Latency Big Data Pipeline on Cloud: Data pipelines are railways for ML wagons. From a data engineering perspective, the desired characteristics of a data pipeline are: accessibility, scalability, efficiency and monitoring. Having robust data pipelines in production enables a company to develop and deploy data-powered solutions that are solving business problems. But it’s not easy. It requires alignment between data engineers, data scientists and the rest of the business. Hard work? Yes, but definitely worth the benefits. (Satish Chandra Gupta @ Towards Data Science)
- AI Expert Roadmap: Technologies and methods to adopt on the way to becoming a data scientist, ML engineer or data engineer. Each stop on the roadmap includes a link so it might be worth bookmarking the website. (AMAI)
- AI ethics groups are repeating one of society’s classic mistakes: The article is raising a question whether AI ethics groups consisting predominantly of people from the US and Europe can set globally acceptable guidelines for ethical use of AI. Well, if we have global ambitions, there needs to be a global discussion of variety of opinions. But pragmatically, we need to start somewhere. Like general ethics started in the ancient Greek. (MIT Technology Review)
The festive season is here. I hope you get a chance to spend it with your loved ones. Even if only digitally due to all the travel restrictions and lockdowns. Stay safe, wherever you are, and I hope you get to put your feet up and have a good read between mince pies and mulled wine.