Lightnews — Scholar-powered news

Past is Prologue: How Conversational Analytics Is Changing Data Work | Towards Data Science

Towards Data Science @towardsdatascience.com · 12h

Do your reports explain *why* something is happening and what to do next? Most don't. This new article by Whitney Marks shows you how conversational analytics can provide not just an answer, but a recommended next step.

The future of reporting will be about encoding the value proposition of a product into prompt design.

How the Rise of Tabular Foundation Models Is Reshaping Data Science | Towards Data Science

Towards Data Science @towardsdatascience.com · 14h

Learn how Tabular Foundation Models (TFMs) generate predictions for new datasets without retraining. Pirmin Lemberger's new article explains this new approach, which uses in-context learning for tabular data.

A turning point for data analysis?

An easy starting point for larger and more complicated Dash dashboards

1 2

Reposted by Towards Data Science

davidi @davidi99.bsky.social · 1d

Plotly Dash — A Structured Framework for a Multi-Page Dashboard | Towards Data Science

Use AI to Supercharge Your Learning on roadmap.sh

Reposted by Towards Data Science

roadmap.sh @roadmapsh.bsky.social · 15h

Stop struggling, start learning anything. ✅ Watch the replay where @kamrify.bsky.social, creator of Roadmap, walks you through our AI Tutor. See the full demo and learn how to get a dedicated personal tutor for any topic you choose.

In this session, I'll walk you through our AI Tutor, share the quick story of how and why we built it, then jump into the demo. You will get the chance to ask questions at the end. Come see how to…

www.youtube.com

2 3

Towards Data Science @towardsdatascience.com · 15h

Learn how to implement the MobileNetV2 architecture from scratch in @PyTorch.org. Muhammad Ardi explains the key concepts from the original paper, including inverted residuals and linear bottlenecks, and provides a complete, step-by-step coding guide.

MobileNetV2 Paper Walkthrough: The Smarter Tiny Giant | Towards Data Science

Understanding and implementing MobileNetV2 with PyTorch — the next generation of MobileNetV1

Data Visualization Explained (Part 2): An Introduction to Visual Variables | Towards Data Science

Towards Data Science @towardsdatascience.com · 17h

Murtaza Ali breaks down the concept of visual variables. The building blocks of effective data visualization, and explains how they shape what we see and understand.

A non-technical and accessible guide to the underlying concept behind visual design: visual encoding channels

Build a Data Dashboard Using HTML, CSS, and JavaScript | Towards Data Science

Towards Data Science @towardsdatascience.com · 19h

For data scientists looking to bring their insights to life, @taupirho.bsky.social shows how to create interactive dashboards without relying on front-end frameworks.

A framework-free guide for Python programmers

Know Your Real Birthday: Astronomical Computation and Geospatial-Temporal Analytics in Python | Towards Data Science

Towards Data Science @towardsdatascience.com · 1d

Learn to solve for a precise astronomical event (the solar return ☀️) using Python. Chinmay Kakatkar's new article discusses how these methods apply to professional domains, including solar energy optimization, space mission planning, and historical event reconstruction.

A hands-on walkthrough using skyfield, timezonefinder, geopy, and pytz, and further practical applications

Eulerian Melodies: Graph Algorithms for Music Composition | Towards Data Science

Towards Data Science @towardsdatascience.com · 1d

Ever wondered how math meets music? Chinmay Kakatkar walks through using graph theory to generate melodies with Python in a fascinating blend of art and algorithms.

Conceptual overview and an end-to-end Python implementation

Towards Data Science @towardsdatascience.com · 1d

🌟 TOMORROW! 🌟 Is speeding up code just shifting the bottleneck to MLOps? 🤔 Join our sister site @TheNewStack.io for this free webinar to learn actionable ways to rethink your testing and deployment workflows for seamless AI integration.

Register free now ➡️ bit.ly/Antithesis-A...

Data Visualization Explained (Part 3): The Role of Color | Towards Data Science

Towards Data Science @towardsdatascience.com · 1d

🎨 Learn to use color with purpose in your data visualizations. Murtaza Ali's new article covers the fundamental principles, from choosing the right color scale for your data type to ensuring your designs are accessible for all viewers.

A simple and powerful guide to using color for more impactful data stories.

Learn how to access vasts amounts of information with your own deep research system

Reposted by Towards Data Science

davidi @davidi99.bsky.social · 3d

How to Build a Powerful Deep Research System | Towards Data Science

How To Build Effective Technical Guardrails for AI Applications | Towards Data Science

Reposted by Towards Data Science

dmitryontech.bsky.social @dmitryontech.bsky.social · 1d

How To Build Effective Technical Guardrails for AI Applications in @towardsdatascience.com

towardsdatascience.com/how-to-build...

Exploring the most practical guardrails to implement at ground level

How I Used ChatGPT to Land My Next Data Science Role | Towards Data Science

Reposted by Towards Data Science

roadmap.sh @roadmapsh.bsky.social · 2d

🚨 Your personal learning coach is here! 🚨 Our AI Tutor on Roadmap includes...

✔️ Floating chat on every roadmap
✔️ Topic guides + code samples
✔️ AI-powered quizzes
✔️ Save progress, resume anytime

Try it 👉 roadmap.sh/ai

2 2

Towards Data Science @towardsdatascience.com · 1d

Prepare for data science product case interviews with a structured, AI-assisted approach. Yu Dong demonstrates how to use ChatGPT to understand business context, brainstorm key metrics, and generate tailored practice questions for specific roles.

Practical AI hacks for every stage of the job search — with real prompts and examples

Python Can Now Call Mojo | Towards Data Science

Towards Data Science @towardsdatascience.com · 1d

@taupirho.bsky.social explains how Python can now call Mojo, offering a powerful way to overcome performance bottlenecks.

Boost your runtimes with lightning-fast Mojo code

How To Build Effective Technical Guardrails for AI Applications | Towards Data Science

Towards Data Science @towardsdatascience.com · 2d

Can one LLM be used to judge another? Yes, and it's a powerful guardrail for validating model outputs at scale. In Nidhin Karunakaran Ponon's debut TDS article, he explains the "LLM as a Judge" approach to replace manual intervention.

Exploring the most practical guardrails to implement at ground level

This Puzzle Shows Just How Far LLMs Have Progressed in a Little Over a Year | Towards Data Science

Towards Data Science @towardsdatascience.com · 2d

Are you aware of the remarkable strides LLMs have made recently?? @taupirho.bsky.social presents a fascinating comparison. Sonnet 4.5 solved a geometric puzzle in seconds, a task that took its predecessor hours longer. This case study underscores the monumental improvements in AI technology.

What took GPT-4o 2 hours to solve, Sonnet 4.5 does in 5 seconds

How to Perform Effective Agentic Context Engineering | Towards Data Science

1 3

Towards Data Science @towardsdatascience.com · 2d

Reduce context bloat and prevent performance degradation in your AI agents. Eivind Kjosbakken's new article includes strategies for shortening an agent's context, including using a dedicated shortening LLM and optimizing for recall before precision.

Learn how to optimize the context of your agents, for powerful agentic performance

Docling: The Document Alchemist | Towards Data Science

Reposted by Towards Data Science

taupirho.bsky.social @taupirho.bsky.social · 3d

PDFs are great for sharing, but a nightmare for data analysis. My new article on @towardsdatascience.com introduces Docling, a Python library that revolutionises PDF processing and OCR. Read it for free using the link below,

towardsdatascience.com/docling-the-...

Why do we still wrestle with documents in 2025? Spend some time in any data-driven organisation, and you’ll encounter a host of PDFs, Word files, PowerPoints, half-scanned images, handwritten notes, a...

This Puzzle Shows Just How Far LLMs Have Progressed in a Little Over a Year | Towards Data Science

1 3

Reposted by Towards Data Science

roadmap.sh @roadmapsh.bsky.social · 3d

Stop copy-pasting #SQL queries and master how to write and optimize them yourself.

This course by Kamran Ahmed will help you build confidence in databases, solve 100+ real challenges, and showcase your skills with a certificate.

Start here ▶️ roadmap.sh/courses/sql

2 3

Reposted by Towards Data Science

taupirho.bsky.social @taupirho.bsky.social · 2d

A tricky geometry puzzle shows just how much leading LLMs have progressed in little over a year. I detail everything in my latest article on the @towardsdatascience.com platform. Are you smart enough to solve the puzzle?

towardsdatascience.com/this-puzzle-...

What took GPT-4o 2 hours to solve, Sonnet 4.5 does in 5 seconds

Plotly Dash — A Structured Framework for a Multi-Page Dashboard | Towards Data Science

Towards Data Science @towardsdatascience.com · 2d

Learn to structure a multi-page @Plotly.com Dash application. Read Michael Clayton's new article which dives into a fully functional repository that moves beyond single-file examples for easier project management.

An easy starting point for larger and more complicated Dash dashboards

Towards Data Science @towardsdatascience.com · 2d

Have you built a project you're proud of? 💻 Our readers are hungry for practical, code-driven insights.

Share your most recent breakthroughs, side projects, or lessons learned. Your clarity and curiosity are what make TDS exceptional.

✍️ bit.ly/TDSContributor