Micro Free Course — 100% Free Learning
All lessons in this module are free to learn. Sign in with Google to save your progress.
Supervised Learning
Decision trees and random forests, gradient boosting (XGBoost/LightGBM/CatBoost), regularization (L1/L2/early stopping), feature engineering for ML, and handling class imbalance.
Module Content
Decision Trees and Random Forests
The model that dominates tabular ML. Trees, then forests, then why bagging works. FIND_VIDEO: search 'decision tree random forest tutorial' — recommended channel: StatQuest. Aim for 11 min or under.
Quiz: How Trees Decide Splits, and Why Forests Help
Practice quiz: Decision trees are intuitive but unstable. Random Forest fixes the instability with bagging + feature subsampling.
Gradient Boosting — XGBoost, LightGBM, CatBoost
The current dominant family for tabular ML. The intuition behind GBM and why the three libraries exist. FIND_VIDEO: search 'gradient boosting xgboost lightgbm explained' — recommended channel: StatQuest / 3Blue1Brown. Aim for 12 min or under.
Quiz: Boosting Intuition and Hyperparameters
Practice quiz: Boosting builds trees that correct each other's mistakes. The few hyperparameters that matter and how to set them.
Regularization — L1, L2, Early Stopping
The three techniques that prevent overfitting. When and how to use each. FIND_VIDEO: search 'regularization L1 L2 dropout machine learning' — recommended channel: StatQuest. Aim for 10 min or under.
Quiz: Picking the Right Regularizer
Practice quiz: L1 vs L2 vs early stopping. Each has a different effect on the model. Pick by what you need.
Feature Engineering and Categorical Encoding
Models are only as good as their features. The encoding choices and transformations that move the needle. FIND_VIDEO: search 'feature engineering categorical encoding tutorial' — recommended channel: Kaggle / Andrew Ng. Aim for 11 min or under.
Quiz: Encoding, Scaling, Interactions
Practice quiz: The feature-prep choices that matter — and the ones that don't. Plus what to do about categoricals.
Class Imbalance — SMOTE, Class Weights, Threshold Adjustment
When 1% of your data is the class you care about. Three strategies for handling it. FIND_VIDEO: search 'class imbalance SMOTE class weight' — recommended channel: StatQuest. Aim for 10 min or under.
Quiz: When and How to Handle Imbalance
Practice quiz: The three main techniques and when each one works. Plus the most common mistake (treating accuracy as the metric).
SUBMISSION: Project 2 — End-to-End Churn Prediction with Cost-Sensitive Loss & ROC/PR Curves
Ship a production-grade churn scorer on 100K telecom subscribers: cost-sensitive logistic regression, calibrated probabilities, ROC/PR analysis, and a cost-optimal calling threshold.
SUBMISSION: Project 3 — Feature Engineering & Preprocessing Pipeline with Scikit-Learn Pipelines
Build a leakage-proof retail demand pipeline: imputation, scaling, categorical encoding, and in-fold target encoding — with a measured leaky-vs-clean gap to prove the discipline.