1 · Handling messy data
Real-world data has gaps, outliers, and class imbalance — three problems every ML pipeline must defuse before model.fit().
●Track 4 · live
EDA & Feature Engineering
What to do before any model.fit() — clean, explore, encode.
Handling missing values, outliers, encoding categoricals, balancing classes with SMOTE, and three full EDA case studies (Wine Quality, Flight Price, Google Play Store). Pandas + matplotlib in the browser.
10 / 12 lessons3 chapters
2 · Categorical encoding
One-hot, ordinal, label, and target encoding — picking the right transformation for nominal vs ordinal features.
3 · EDA case studies
Three end-to-end EDA + feature engineering walkthroughs — wine quality, flight prices, Google Play Store cleaning and exploration.