Method and implementation of machine learning software engineering
本书是一本面向机器学习的进阶读者的机器学习工程实战宝典。 作者将软件工程方法引入到机器学习的工程实践中，两者的亲密接触和融合定会给读者带来新的体会和视野。 作者融合了自己10年丰富的工程实践经验，详细阐述机器学习核心概念、原理和实现，并提供了数据分析和处理、特征选择、模型调参和大规模模型上线系统架构等多个高质量源码包。
全书共16章，分为4个部分：
工程基础篇（1～3）： 介绍了机器学习和软件工程的融合，涉及理论、方法、工程化的数据科学环境和数据准备；
机器学习基础篇（4～5）： 讲述了机器学习建模流程、核心概念，数据分析方法。
特征篇（6～8）： 详细介绍了多种特征离散化方法和实现、特征自动衍生工具和自动化的特征选择原理与实现；
模型篇（9～16）： 深入的讲述了线性模型、树模型和集成模型原理与模型剖析；基于模型基础，进一步讲述了模型调参方法、自动调参原理与实现、模型评估和不同模型（白盒，黑盒）解释原理与实现；模型上线之模型即服务一章提供了5种工程化的模型上线方法；最后以模型监控一章结束机器学习项目流程的最后一环。
This book is a practical collection of machine learning engineering for advanced readers of machine learning. The author introduces software engineering methods into the engineering practice of machine learning, and the close contact and integration of the two will definitely bring new experience and vision to readers. The author integrates his 10 years of rich engineering practice experience, elaborates on the core concepts, principles and implementation of machine learning, and provides multiple high-quality source codes such as data analysis and processing, feature selection, model tuning, and large-scale model deployment system architecture package.
The book consists of 16 chapters, divided into 4 parts:
Engineering Fundamentals (1～3): Introduces the integration of machine learning and software engineering, involving theories, methods, engineering data science environment and data preparation;
Machine Learning Fundamentals (4~5): Describes the machine learning modeling process, core concepts, and data analysis methods.
Features (6～8): Introduced in detail the methods and implementation of various feature discretization, feature automatic derivative tools and automatic feature selection principle and implementation;
Model section (9-16): In-depth description of linear model, tree model and integrated model principle and model analysis; based on the model foundation, further describes the model parameter adjustment method, automatic parameter adjustment principle and realization, model evaluation and different models ( White box, black box) explain the principle and implementation; the model-as-a-service chapter of model deployment provides 5 kinds of methods; finally, the chapter model monitoring ends the last step of the machine learning project process.