Volume II is where the work starts to look like the job. You will train a model in the first few pages — and then spend the rest of the chapter earning the right to trust it: honest validation, why 97% accuracy on an imbalanced dataset is a lie, and the bias-variance tradeoff as something you can see rather than recite. From linear and logistic regression to decision trees, random forests and gradient boosting, you build models that generalise and explain them to people who don't write code. Then the unlabelled half of real data: k-means, hierarchical clustering and DBSCAN, PCA and t-SNE, anomaly detection, and segmentation a marketing team can actually act on. Time series and recommender systems get the careful treatment they rarely receive, and a full chapter on product analytics teaches the skill most courses skip — deciding whether the prediction was worth making at all.

Volume II of the "Data Science, End to End — Essentials" series by Archit Sharma.