Python Machine Learning, 2nd Edition (Chinese Translation)

This book is the Chinese translation of Python Machine Learning (2nd Edition), translated by Chen Bin and published in the China Machine Press "Data Science and Engineering Technology Series." It comprises 16 chapters built around two threads—principles and code: it covers the necessary details of machine learning concepts while explaining why each algorithm is designed as it is, how to use it, and where things commonly go wrong. It opens with the basic concepts of machine learning and predictive analytics, explains why Python has become the mainstream language of data science, and covers setting up a development environment and preparing data. It then demonstrates implementing classification algorithms from scratch using examples such as the perceptron and the adaptive linear neuron, before moving on to calling various classifiers with scikit-learn. The middle section focuses on data preprocessing, feature selection and dimensionality reduction (including principal component analysis and linear discriminant analysis), model evaluation and hyperparameter tuning, and ensemble learning, helping readers build a complete workflow for judging whether a model is reliable. The latter half turns to applications and deep learning: sentiment analysis is used to demonstrate text feature extraction and classification and to embed a trained model in a web application; regression analysis predicts continuous target variables; clustering methods handle unlabeled data; and the book closes by implementing a multilayer neural network from scratch and then building and training networks with TensorFlow, covering convolutional neural networks for images and recurrent neural networks for sequence data. The 2nd edition is substantially rewritten and expanded from the 1st, introducing tools such as Keras and TensorFlow and taking deep learning from principles through to engineering practice. The companion code is open-sourced on GitHub (rasbt/python-machine-learning-book-2nd-edition), so readers can run it as they read.