Intro to Statistical Learning — the gentle on-ramp

Ten interactive modules covering the ground of the classic introductory course (the topics of James, Witten, Hastie & Tibshirani's An Introduction to Statistical Learning), rebuilt as plain-English lessons with a live visual lab in every one — no math background assumed, every concept something you manipulate rather than memorize. Fit a regression line with your own hands, slide a classification threshold and watch the errors trade, cut a dendrogram and see the clusters reorganize. When you finish, the 18-module advanced track continues the same story with the deeper machinery — shrinkage paths, boosting, neural nets, graphical models and the p ≫ N world.

  1. 🔎
    Module 1Available

    Learning from data

    Signal vs noise, prediction vs inference — and a three-dataset mystery where one dataset hides nothing at all.

    🎛 Three-dataset mystery

  2. 🎚️
    Module 2Available

    The bias-variance tradeoff

    Refit the same model on 30 parallel-universe training sets and watch rigidity and wobble trade places.

    🎛 Parallel-universe lab

  3. 📐
    Module 3Available

    Linear regression by hand

    Drag your own line, watch the residuals, race the least-squares answer — RSS and R² from first principles.

    🎛 Fit-the-line game

  4. 🚨
    Module 4Available

    Classification & thresholds

    The model outputs probabilities; you choose where to cut. Confusion matrices, base rates and the ROC curve.

    🎛 Threshold & ROC lab

  5. 🔁
    Module 5Available

    Resampling methods

    Validation split vs 5-fold vs 10-fold vs LOOCV — fifteen reruns each, so you can see which estimate you could trust.

    🎛 Estimate-spread lab

  6. 🗂️
    Module 6Available

    Model selection

    Forward stepwise picks the real features before the decoys — then keeps going. R² vs adjusted R² vs test error.

    🎛 Stepwise selection lab

  7. 〰️
    Module 7Available

    Beyond linearity

    Bin the axis or raise the degree: two escapes from the straight line with mirror-image failure modes.

    🎛 Steps vs polynomial lab

  8. 🌳
    Module 8Available

    Tree-based methods

    Watch a classification tree carve the plane into rectangles — Gini splits, stair-step boundaries, memorization boxes.

    🎛 2D partition lab

  9. Module 9Available

    SVMs & kernels

    A blob inside a ring defeats every straight line — until a radial kernel teaches the machine to draw circles.

    🎛 Radial-kernel lab

  10. 🌿
    Module 10Available

    Hierarchical clustering

    The dendrogram: every possible K in one tree, linkage personalities, and a cut line you slide yourself.

    🎛 Dendrogram lab

How the two tracks fit together

This track builds the intuitions — one idea per module, one visual per idea. The advanced track revisits the same territory with the practitioner's machinery and extends it to boosting, neural networks, SVM margins, graphical models and high-dimensional problems. Corresponding modules cross-link both ways, so you can drop down for intuition or climb up for depth at any point.

Original interactive lessons and examples; all labs run seeded synthetic data in your browser. Educational analysis, not investment advice.