Classical Machine Learning: Supervised Learning Foundations

Master the mathematical foundations and practical implementation of classical supervised learning algorithms. Build intuition through interactive visualizations while understanding the theory that powers modern ML.

Learning Objectives

Understand the mathematical foundations of supervised learning
Master linear models: regression and classification
Build intuition for decision trees and their ensembles
Implement Support Vector Machines from scratch
Master model evaluation and validation techniques
Understand bias-variance tradeoff and regularization
Learn feature engineering and selection strategies
Apply ML algorithms to real-world problems

Интерактивные инструменты в этом курсе

Осваивайте концепции через практическое изучение

Loading tool...

Click to InteractExplore

⚖️

Bias-Variance Tradeoff Explorer

interactive

Interactive visualization of the bias-variance tradeoff and model complexity effects

Explore Full Tool

Loading tool...

Click to InteractExplore

🎯

ML Model Trainer

interactive

Train machine learning models interactively with real-time visualization

Explore Full Tool

Loading tool...

Click to InteractExplore

🎨

Decision Boundary Visualizer

visualization

2D visualization of classification decision boundaries

Explore Full Tool

Смотреть все инструменты

Course Content

Mathematical Foundations of Machine Learning

Build the mathematical framework for understanding ML: linear algebra essentials, probability theory, optimization basics, and the learning problem formulation.

60 min

The Supervised Learning Framework

Understand the core concepts: training vs testing, loss functions, empirical risk minimization, and the bias-variance decomposition.

60 min

Linear Regression: From Theory to Practice

Master linear regression from first principles: least squares, maximum likelihood, gradient descent optimization, and geometric interpretation.

60 min

Logistic Regression and Classification

Extend to classification: sigmoid function, maximum likelihood for classification, decision boundaries, and multi-class strategies.

60 min

Regularization: L1, L2, and Elastic Net

Prevent overfitting with regularization: Ridge, Lasso, and Elastic Net. Understand the geometry and sparsity-inducing properties.

60 min

Decision Trees: Intuition and Implementation

Build decision trees from scratch: entropy, information gain, tree construction algorithms, and pruning strategies.

75 min

Random Forests and Bagging

Ensemble power through bagging: bootstrap aggregating, random forests, feature randomness, and out-of-bag evaluation.

60 min

Gradient Boosting: From AdaBoost to XGBoost

Master boosting algorithms: AdaBoost, Gradient Boosting, and modern implementations like XGBoost and LightGBM.

75 min

Support Vector Machines: Linear Case

Understand SVM geometry: maximum margin classification, support vectors, primal and dual formulations, and soft margins.

75 min

Kernel Methods and Non-linear SVMs

Transform to higher dimensions: kernel trick, RBF/polynomial/sigmoid kernels, and the connection to neural networks.

75 min

Evaluation Metrics: Beyond Accuracy

Master evaluation metrics: precision/recall, F1-score, ROC curves, AUC, and choosing the right metric for your problem.

60 min

Cross-Validation and Model Selection

Robust model evaluation: k-fold CV, stratified sampling, time series CV, nested CV for hyperparameter tuning.

60 min

The Art of Feature Engineering

Transform raw data into ML-ready features: scaling, encoding, polynomial features, and domain-specific transformations.

60 min

Feature Selection and Dimensionality Reduction

Select the best features: filter/wrapper/embedded methods, and reduce dimensions with PCA and feature importance.

60 min

End-to-End ML Project: From Data to Deployment

Apply everything in a real project: data exploration, feature engineering, model selection, evaluation, and deployment preparation.

90 min