New tools and packages come and go, but the basic grammar of data visualization remains incredibly resilient to trends: at the end of the day, we still need to combine lines, color, and text in an effective way to tell the story of our data.
That does not mean, alas, that it’s always easy or straightforward to find the right approach for visualizing data-based insights. Err too much on the side of simplicity, and our charts might look boring—or even dated. Add too many splashy touches, and we risk overwhelming and distracting our readers, customers, and stakeholders.
When there’s no one-size-fits-all solution, and where every project calls for a thoughtfully tailored approach, the best thing data professionals can do is continuously grow their visualization toolkit, and learn—via trial and error—what works best in different contexts. Our recommended posts this week will help you along that journey: they offer concrete ideas, and stress the importance of matching visuals to the underlying message you want to convey.
Beyond Bar Charts: Data with Sankey, Circular Packing, and Network Graphs
There’s nothing wrong with a good, clean bar chart, of course, but there are times when the story we’re telling is just too complex and multilayered to fit neatly within one. Maham Haroon offers a detailed guide to three elegant alternatives you might consider next time you’re stuck: Sankey diagrams, circular packing, and network graphs.Creating Animation to Show Four Centroid-Based Clustering Algorithms using Python and Sklearn
It can be a challenge to present the results of a clustering process in an engaging way. Boriharn K proposes a neat idea for those times when a series of plots just won’t cut it: animating your work. As Boriharn points out, it “can be useful in showing how each algorithm works and monitoring the change in the process.”Make Beautiful (and Useful) Spaghetti Plots with Python
Maybe it’s something about food-themed visualizations (pie charts, anyone?), but spaghetti plots tend to be the object of occasional derision. Lee Vaughan demonstrates how they can nevertheless be used effectively, and leans on the timely example of climate-change data to explain how you can create them.Photo by Matt Briney on Unsplash
From Rubik’s Cubes to LLMs, we’ve published some wonderful articles on other topics, too—here’s a selection of some of our recent highlights:
Parul Pandey makes a powerful argument for the importance of culture when it comes to organizations’ ability to adopt responsible AI practices.If you love history and math (and/or the history of math), you absolutely shouldn’t miss Sachin Date’s deep dive on the Markov and Bienaymé–Chebyshev inequalities.What should a machine learning project roadmap look like? Why should you even have one? Heather Couture’s recent overview offers clear and actionable answers to these questions.Expand your deep learning toolkit and explore normalization as an efficient optimization technique during model training— Thao Vu’s guide is a great place to start.There’s always a new algorithm to learn about! Kay Jan Wong’s explainer covers the Reingold-Tilford algorithm and includes a thorough walkthrough of its inner workings.For his debut TDS article, Eduardo Testé focuses on probability in the context of a planning problem with a colossal state space and only one solution: the Rubik’s Cube.From retrieval-augmented generation (RAG) to parameter efficient fine-tuning (PEFT), Maarten Grootendorst presents several helpful methods to improve the performance of your large language model.
Thank you for supporting our authors’ work! If you enjoy the articles you read on TDS, consider becoming a Medium member — it unlocks our entire archive (and every other post on Medium, too).
How to Choose the Right Data Visualization Strategy for Your Project was originally published in Towards Data Science on Medium, where people are continuing the conversation by highlighting and responding to this story.
