Foundations
Data Fundamentals | Visual Thinking | Model Evaluation
Foundations
Data Fundamentals | Visual Thinking | Model Evaluation
Free, open-source materials for learning data science by doing. Every exercise runs in the browser — no installation, no setup, no excuses.
These aren’t lectures. Each module is built around working code, real output, and problems designed to break in useful ways.
Try Python
Browser-based fundamentals via Pyodide
Data types, variables, functions, iteration, and flow control. Click run, see output, change something, run again. Zero setup.
Best for: complete beginners or anyone who wants to verify their Python intuition before moving on.
Explore Data
Working with real datasets
Loading, inspecting, filtering, and visualizing data with pandas and plotly. JupyterLite-backed notebooks that run client-side with persistent state across cells.
Best for: people who know basic Python and want to work with actual data.
Build Models
ML workflows with full environments
Classification pipelines, model evaluation, and sampling methods. Binder-backed notebooks with scikit-learn and full package access — real environments for real workflows.
Best for: people comfortable with data manipulation who want hands-on ML practice.
Why Three Tracks?
Each track uses a different execution environment matched to its content. Try Python uses Pyodide because zero friction matters when you’re learning syntax. Explore Data uses JupyterLite because multi-step analysis needs persistent state. Build Models uses Binder because ML workflows need real packages and real compute.
This is intentional. Read more about the design decisions behind these choices →
Open Source
All content, code, and platform configuration for Pixel Process is available on GitHub. Use it, fork it, improve it.