Machine learning, a software technique that learns from experience, has made a comeback in recent years. In the early days when computers were born, the concept of machine learning had already appeared, and various theories were unconstrained and failed to be popularized due to the cost of computing. With the proliferation of computing devices and the increasing use of machine learning in everyday life, one can say that its success has become commonplace. New applications are springing up, many powered by machine learning.

In this book, we will look at some machine learning models and algorithms. We will introduce some commonly used methods for evaluating the effects of machine learning tasks and models. These models and algorithms are implemented through the popular Python machine learning library SciKit-Learn, which contains many machine learning models and algorithms, and each API is easy to use.

The book features:

  • The content is easy to understand. This book requires only basic programming and mathematics
  • Practical case. The cases in this book are easy to use and readers can adapt to solve their own problems.

Chapter 1, Fundamentals of machine learning, defines machine learning as a process of program research and design that improves work performance through learning experience. The rest of the chapters are based on this definition, and the machine learning models introduced in each subsequent chapter are based on this idea to solve tasks and evaluate results.

Chapter 2, Linear regression, introduces linear regression models, in which explanatory variables and model parameters are correlated with continuous response variables. This chapter introduces the definition of cost function and obtains the optimal model by solving model parameters by least square method.

Chapter 3, feature extraction and processing, introduces the feature extraction and processing methods of common machine learning objects such as text, image and classification variables.

Chapter 4, from linear regression to logistic regression, introduces how the generalized linear regression model solves classification tasks. Combining logistic regression model with feature extraction technology, a spam SMS classifier is implemented.

Chapter 5, decision tree — nonlinear regression and classification, introduces a nonlinear model of regression and classification — decision tree. A web advertising image blocker is implemented by decision tree integration method.

Chapter 6, K-means clustering, introduces the k-means clustering algorithm of unsupervised learning, and combines it with logistic regression to realize a photo classifier.

Chapter 7, dimension reduction with PCA, introduces another unsupervised learning task – dimension reduction. We use principal component analysis to realize visualization of high dimensional data and build a face recognition machine.

Chapter 8, Perceptron, introduces a real-time, binary classifier called perceptron. The next two chapters develop in response to the shortcomings of perceptrons.

Chapter 9, from perceptron to support vector machine, introduces support vector machine, which is an effective nonlinear regression and classification model. We use support vector machines to identify letters in street view photos.

Chapter 10, from perceptron to artificial neural network, introduces artificial neural network, which is a powerful and effective nonlinear regression and classification model. We use artificial neural networks to recognize handwritten numbers.