Overview
An interactive page that runs a Jupyter notebook cell by cell, walking through ML classification with hands-on examples. Covers the full pipeline from data loading to evaluation, with four extensions for the more advanced classifiers.
The Problem
Static docs don't capture how ML actually feels to do. You need to see how each step changes the output, not just read what the code is supposed to do.
The Approach
Built an interactive HTML page that simulates the notebook with auto-typed code, live syntax highlighting, and context for each cell. Every cell has a helper panel explaining the concept, the expected output, and where it sits in the theory. NCM, LDA, QDA, and Mahalanobis distance classifiers all implemented from scratch.
Outcome
A self-contained teaching resource that makes the equal vs unequal covariance distinction concrete. Dataset A reaches 100% accuracy with NCM, Dataset B drops to 98.5%, and the contrast shows clearly when a linear boundary is enough versus when you need a quadratic one.
