If you want to get serious in learning machine learning, these books are the classics that cannot go wrong.

### Pattern Recognition and Machine Learning by Christopher Bishop, 2006

- Free PDF from Microsfot
- Known as the PRML, this 700 page book is for the advanced undergraduates or beginning graduate students.
- You definitely need some fundamental in math
- Probability theory (ie. Expectation, Conditional Probability)
- Linear Algebra (ie. vector and matrixes)
- A little bit of calculus

- The book was published in 2006, so it does not have all the new and shiny deep neural network based model, but it provides a very solid understanding in the Bayesian view on machine learning that will benefit you even today.
- It covers topics in:
- Linear Regression/Classification
- Neural Network
- Kernel Method
- Graphical Model
- Mixture Model and Expectation Maximization
- Approximate Inference
- Sampling Methods
- Hidden Markov Model

- Don’t read this book on Kindle

### An Introduction to Statistical Learning with Applications in R by Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani, 2014

- Free PDF
- This book is specialized for 2 things:
- It covers all the traditional machine learning methods
- It uses R

- The tradition ML models are included:
- Linear Regression
- Logistic Regression
- LDA (linear discriminant analysis), QDS
- K-Nearest Neighbor
- Regularization:
- Ridge Regression
- Lasso Regression

- Principal Component Analysis (PCA)
- Polynomial Regression
- Tree-based Methods
- Decision Tree
- Bagging Trees
- Random Forest
- Boosted Trees

- Support Vector Machine (SVM)
- Unsupervised learning: K-mean Clustering

- This book is a great book for everything except Deep Learning because R is not so specialized in Deep Learning frameworks.

### Machine Learning: A Probabilistic Perspective by Kevin Murphy, 2012

- This book is 1000 pages and quite a comprehensive ML that cover so many topics.
- It is really quite a text book on its own. It will gives you the knowledge from basic probability theory to statistical thinking.
- It does not just give you an overview all the ML models but it always starts from the fundamental concept and connecting them to the model.
- It includes basically any class of model that you will encounter.
- However, this book is quite overwhelming with all the details to just study in one course. It’s more suitable to be taught in multiple course series.
- It covers both traditional and deep learning methods.
- I could classify this book as a advanced book for the experts to reference to.

### Deep Learning by Ian Goodfellow, Yoshua Bengio and Aaron Courville, 2016

- Website: https://www.deeplearningbook.org/
- This is “the deep learning” book by Ian Goodfellow.
- This is a bible for many ML practitioners.
- However, this isn’t a great book for self study according to many reviewers on this book. Sometimes, great researchers might not always be great teachers.

### The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Trevor Hastie, Robert Tibshirani and Jerome Friedman

- Free PDF
- Another very comprehensive book with more rigorous math.
- It goes into depth on the derivation on many algorithms.
- Definitely a great reference for those who want to study applied statistics.
- It’s cover topics in both traditional and modern machine learning:
- Linear regression with Lasso (the author of the book is one of the creator of Lasso)
- Tree Models – boosting and additive trees
- Neural Network
- Support Vector Machines
- K-Nearest Neighbor
- Unsupervised learning
- Spectral clustering
- kernel PCA
- Matrix factorization
- Google page rank

- Random forests
- Ensemble Learning
- Graphical models

Disclaimer

Many free links are provided for the book but often reading the book in printed copies is much easier to digest, so I am promoting these books as marking for these books. Please buy them if they are useful for your career.