Want to learn about the world of Bayesian statistics and better understand the difference between Bayesian and frequentist methods? Psychologist and consultant Hannah Roos wrote the resource you’ve been looking for. After introducing the necessary terms, Hannah dives into a brilliant case study (involving weddings and weather!) that makes these concepts come alive. If you want to know when, why, and how you should use Bayesian statistics, this article is for you.
Advances in likelihood-free inference and meta-learning made Arthur Pesah (PhD student in quantum computing at UCL) and Antoine Wehenkel (PhD Student at Belgium’s National Fund for Scientific Research) wonder: “Can we build a machine that takes a tweakable simulator and real data as input, and returns the version of the simulator that fits best some real data?” Read this eye-opening article to find out what they discovered. It offers multiple GIFs and graphs to ease your exploration of these complex concepts.
Now that the machine learning community has recognized the need to take algorithmic bias seriously, it’s time to explore potential ways to mitigate it — including algorithmic solutions. In this article, former DeepMind Research Engineering intern Joyce Xu asks, “how, algorithmically, can we ensure that the models we build are not reflecting and magnifying human biases in data?” From adversarial de-biasing to distributionally robust optimization, Joyce dives into several key potential fixes. A great long read!
“Decision intelligence is the discipline of turning information into better actions at any scale.”
In this comprehensive 14-minute read, Cassie Kozyrkov, Head of Decision Intelligence at Google, invites us to learn more about her field of expertise. After introducing important terminology and concepts — from “what’s a decision” to “what’s the difference between making a calculation versus making a decision” — Cassie digs deeper into this fascinating topic and demonstrates how relevant it is for the work of all data scientists.
Distance measures are ubiquitous in data science and machine learning. Many algorithms rely on them, and “knowing when to use which distance measure can help you go from a poor classifier to an accurate model.” In this article, Maarten Grootendorst does a fantastic job explaining the most common ones. For each of the nine distance measures, Maarten describes how they work, offers several use cases, and discusses their potential drawbacks.
Since we frequently receive questions about how to contribute to TDS, I’ve gathered some thoughts that I hope will help those interested in getting started.
First and foremost, we have put a lot of effort into our Write for Towards Data Science article. If you haven’t looked at…
We launched Towards Data Science about three years ago, and we’re still truly astonished to see the engagement we get from so many of you. Today, we’re a team of four editors and twenty editorial associates, and since we’ve been receiving questions from you, we decided to do our best…
Science is a quest to reach good explanations about the world. As David Deutsch pointed out in his book, The Beginning of Infinity, a good explanation is clear, precise and hard to vary.
Data science is a new scientific field that thrives to extract meaning from data and improve understanding…