Happy new year, everyone! The beginning of January is a good moment to take bold steps, make ambitious plans, and jump headfirst into new challenges and interests. To help you with all of that, our first Variable edition of the year features 10 recent posts that we found particularly energizing and useful. Let’s get right to it!
- Are you just getting started in Data Science? Is becoming a full-time data scientist a key item on your 2022 to-do list but you’re not sure where to start? To help you with your first steps, we’ve compiled some of our most popular resources for beginners and learners in our latest Monthly Edition.
- Give your communication skills a boost. Data scientists often bridge gaps between technical and non-technical stakeholders, which, according to Rose Day, is precisely why they should focus on effective, context-driven communication.
- Make the most of your unlabeled data. If you’re up for a hearty deep dive that combines theory with practical applications, look no further than Betty LD‘s recent article on uncertainty, smart sampling, and getting the most value out of your available data.
- Catch up with recent trends in Graph Machine Learning. As one of the most dynamic areas within machine-learning research, it’s easy to get lost amid all the newest developments in graph ML. Michael Galkin is here to the rescue with a comprehensive, robust overview of the current state of the field.
- What would more "expressive" GNNs look like? Speaking of graph ML, if you’d like to zoom in on a more specific topic within this growing subfield, don’t miss Michael Bronstein‘s latest article, which covers several recent approaches to the application of graph neural networks (GNNs) on a collection of subgraphs.
- Explore the theory underpinning the distillation of large language models like BERT. LLMs have been all the rage for a while now, but training them is extremely resource-heavy. In his latest post, Remi Ouazan Reboul walks us through the distillation techniques that power DistilBERT and alleviate some of the time and costs associated with BERT-like models.
- Stand out from the crowd during your job search. Is finding a new or better data science role on your roadmap for the next few months? Emma Ding recently shared her handy list of dos and don’ts for the product-case interview stage, which you’ll want to read now (and save for future reference).
- Add another visualization resource to your toolkit. If you’re looking for something a bit more hands-on to start off your year, you can’t go wrong with Parul Pandey‘s latest tutorial, which explains how to visualize decision trees and interpret your model using open-source package Pybaobabdt.
- Up your algorithm-powered prediction game. Still in the realm of hands-on learning, Christophe Brown‘s fresh-off-the-press post introduces the support-vector machines algorithm, and shows how you can use it to (attempt to) predict the results of NBA games while they’re still being played.
- Grow your understanding of ML interpretability. Model explainability will continue to be a crucial element in the work of data scientists and ML practitioners in 2022; reading Dr. Robert Kübler‘s newest article will help you build feedforward neural networks that are interpretable by design, using PyTorch.
Thank you for joining us again after our brief holiday pause – we hope your year is off to a good start!
Would joining TDS make it even better? If you’re passionate about data science and good writing, check out our current job opening: we’re looking to hire a fully remote, Canada-based Senior Editor, and would love to hear from you if you think you’d be a good fit for the role.
Until the next Variable, TDS Editors
