Machine Learning A-Z: Python, Statistics & Real Projects

Most machine learning books either drown you in mathematical formalism or hand you library calls with no explanation of what they do. This one does neither.

Machine Learning A-Z is built on a simple conviction: you need both the intuition and the implementation. Every concept is developed from first principles, and every concept is then written out in working Python — not pseudocode, not fragments, but code you can run.

Six parts, twenty-three chapters, four appendices:

• Foundations — what machine learning actually is and how it differs from traditional programming, a reproducible Python environment, and the statistics and probability you genuinely need
• Supervised learning — linear and logistic regression, decision trees and random forests, support vector machines, neural networks and deep learning, gradient boosting with XGBoost, LightGBM and CatBoost, plus k-nearest neighbours and naive Bayes
• Unsupervised learning — k-means, hierarchical clustering and DBSCAN, dimensionality reduction with PCA, t-SNE and UMAP, and autoencoders
• Reinforcement learning — fundamentals, Q-learning and deep Q-networks, and policy gradient methods
• Four complete projects — house price prediction, customer churn prediction, customer segmentation, and a stock trading RL agent, each carried end to end
• Practice and frontier — model evaluation, validation and selection, and a chapter on transformers and large language models

Appendices cover Python environment setup, mathematical notation, further reading, and exercises for every chapter.

Written for practitioners who are comfortable with Python and have a passing familiarity with the mathematics. You do not need a PhD, and you do not need to have taken a formal machine learning course.

Code targets Python 3.12 with scikit-learn 1.7, NumPy 2.2, Pandas 2.3, PyTorch 2.7, TensorFlow 2.19, XGBoost 3.0, LightGBM 4.6 and Gymnasium 1.1. The book notes where later releases have since shipped.