Handbook of Machine Learning for Social Sciences

This handbook is a collaborative work with Angel Hwang and Remy Stewart designed for researchers in social sciences. This handbook is supported by the Cornell Center for Social Sciences and corresponding to the Machine Learning Workshop Series. We expect that the audience will have basic knowledge of Python and have at least taken a course in calculus and statistics. This handbook aims to offer you practical experience with machine learning models.

In this handbook, we are not attempting to teach math behind machine learning. When we walk you through the basic concepts, we use the minimum amount of equations. However, it is essential to understand statistical inference in order to use machine learning in your research better and be able to discuss the advantages and flaws of the application of machine learning models. Therefore, in each chapter (and each section in each chapter), we provide you with a list of resources in case you are interested in digging into certain concepts and/or models.

Chapter 1 introduces machine learning, including the overview of machine learning, different types of machine learning models, and the application of machine learning for social sciences. In Chapter 2, we cover Python basics to prepare you for data preprocessing steps (data visulzation and feature engineering) before delving into machine learning tasks. We discuss classification in Chapter 3 and regression in Chapter 4 by introducing the basic models using simulation data and walking you through a case study using the real research datasets. Chapter 5 illustrates the unsupervised learning models and offers you hands-on experience on clustering. Chapter 6 discusses model inference, including prediction and causal inference after training the machine learning models. Finally, we present a glossary of the key concepts illustrated in the handbook and cheatsheets for supervised and unsupervised learning in Chapter 7.

See the Table of Contents with link to each chapter.