Machine Learning Fundamentals

Mission. Classical ML done with the rigor of someone who understands both the math and the engineering.

Scope

Frame a problem as regression, classification, or clustering and justify the choice
Build an evaluation scheme that won't leak or fool itself
Diagnose a struggling model (bias/variance, leakage, drift) systematically

This track is scaffolded and waiting for content. See the curriculum dashboard for the roadmap and progress across all tracks.

In review — 1 draft awaiting publication

Wave 1 — first vertical slice (production-ready) — 26 lessons

The Supervised Workflow concept · beginner · M Data to deployment: the shape of a supervised learning project.
Train, Validation, Test concept · beginner · M The split discipline that keeps your model honest about the future.
Cross-Validation concept · intermediate · M K-fold, stratification, and group leakage.
Data Leakage concept · intermediate · M The silent killer: how information from the future inflates your scores.

Linear Regression as a Model concept · beginner · M Least squares as a model, not just a formula.
Logistic Regression as a Classifier concept · intermediate · M Probabilities, decision boundaries, and calibration.
k-Nearest Neighbors concept · beginner · S The simplest classifier, and what its failures teach.
Decision Trees concept · intermediate · M Splits, impurity, and the overfitting built into trees.
Random Forests concept · intermediate · M Bagging and feature randomness: variance reduction made practical.
Gradient Boosting concept · intermediate · M Sequential error correction, and why it wins on tabular data.
Support Vector Machines concept · advanced · M Margins, kernels, and the geometry of the separator.

k-Means Clustering concept · intermediate · M Assumptions, failure modes, and choosing k honestly.
DBSCAN concept · intermediate · S Density-based clusters without choosing k.
Principal Component Analysis concept · intermediate · M Compression, decorrelation, and interpretation of components.

Regression Metrics concept · beginner · S MSE, MAE, R-squared: what each metric forgives and punishes.
Classification Metrics concept · intermediate · M Precision, recall, F1: choosing the metric that matches the cost.
ROC, Thresholds, Calibration concept · intermediate · M Curves, threshold choice, and whether to trust the probabilities.
Regularization concept · intermediate · M Ridge, lasso, and the bias-variance dial.
Hyperparameter Tuning concept · intermediate · M Grid, random, and Bayesian search without test-set leakage.
Ensembles and Stacking concept · advanced · M Blending models: when diversity beats strength.
Imbalanced Classes concept · intermediate · M When 99 percent accuracy means nothing: resampling, weighting, thresholds.

Feature Selection concept · intermediate · M Choosing fewer, better features without p-hacking your way there.
Model Interpretability concept · intermediate · M Permutation importance and SHAP: interrogating what the model learned.
scikit-learn Pipelines concept · intermediate · M Pipelines and ColumnTransformer: preprocessing without leakage.
Model Persistence and Serving concept · intermediate · M Serializing models and serving predictions behind a function.

Case Study: End-to-End Classification project · intermediate · L A real dataset, honest evaluation, and a served prediction.