Every so often you stumble across a handful of sites that quietly become part of your regular workflow. These nine have earned a permanent spot in my bookmarks bar — a mix of dataset repositories, visual explainers, and tools that make abstract statistics and ML concepts click. Bookmarking them individually always felt scattered, so here they all are in one place, grouped by what they’re actually good for.
Finding Data
1. Google Dataset Search
Exactly what it sounds like — a search engine indexed specifically for datasets across the web, from government agencies to academic repositories. Useful as a first stop when you have a project idea but no data to back it yet.
2. UCI Machine Learning Repository
The old reliable. Hundreds of clean, well-documented datasets that have been the training ground for ML students and researchers for decades. If you want a dataset with a known, well-understood shape to practice a new technique on, this is where to look first.
3. Data Is Plural
Jeremy Singer-Vine’s newsletter and searchable archive of interesting, often obscure datasets — everything from historical shipwrecks to grocery store closures. Less “clean training data,” more “rabbit hole you didn’t know you needed.”
Understanding Concepts Visually
4. Seeing Theory
A Brown University project that turns probability and statistics into interactive visualizations. If a stats concept has ever felt abstract on paper, this is often the fastest way to make it click.
5. Setosa.io
A small but excellent collection of visual explanations for concepts like Markov chains, ROC curves, and eigenvectors. Built by Victor Powell, these are the kind of interactive diagrams that explain in thirty seconds what a textbook paragraph takes three pages to do.
6. Distill.pub
A now-archived but still widely referenced journal for machine learning research, published as interactive visual essays instead of static PDFs. The feature visualization and neural network interpretability pieces in particular are still some of the best explainers on the topic anywhere.
7. Visual Narrative Flow — Exploratory Study
An academic research demo (from a EuroVis paper on data storytelling) comparing “stepper” vs. “scroller” formats for walking a reader through a data narrative. Worth a look less for the dataset itself and more as a case study in how interactive design choices change the way people absorb data stories.
8. OpenIntro Statistics
A completely free, peer-reviewed introductory statistics textbook. If you want the foundational theory without paying textbook prices, this is a genuinely solid option — used in real college courses, not just a marketing funnel.
Building and Presenting
9. Tableau
The industry-standard drag-and-drop visualization tool. Even if you’re primarily a code-first person, it’s worth knowing your way around Tableau since so many teams and job postings expect it.
Nine very different tools, but they all solve the same underlying problem: making data less abstract, whether that’s finding it, understanding it, or explaining it to someone else.