Source: Xiangzhang Economics Academic Circle
On the morning of July 21, 2024, Professor Guo Feng, a doctoral supervisor and head of the Investment Department at the School of Public Economics and Management, Shanghai University of Finance and Economics, was invited to give a wonderful lecture titled “Fundamentals of Machine Learning and Its Economic Applications: Three Case Studies” for the seventh session of the “Xiangzhang Green Seedling Program”.
Guest Introduction
Guo Feng holds a PhD in Economics from Fudan University and a postdoctoral degree in Finance from Peking University. He is currently a professor, doctoral supervisor, and head of the Investment Department at the School of Public Economics and Management, Shanghai University of Finance and Economics. He is a Youth Changjiang Scholar of the Ministry of Education, Executive Director of the Laboratory of Digital and Intelligent Governance at Shanghai University of Finance and Economics, Vice Dean of the Fuguo ESG Research Institute at Shanghai University of Finance and Economics, and a senior researcher at the Digital Finance Research Center of Peking University. He also serves as Vice Chairman of the ESG Committee of the China Enterprise Management Research Association, Executive Director of the New Economy and New Employment Committee of the Labor Economics Society, and a director of the China Society of Quantitative Economics. His research areas mainly include digital economy and digital finance, machine learning and big data analysis, and public economics. He has published over 50 papers in Chinese and English journals such as Economic Research, Management World, Economic Quarterly, and Journal of Economic Behavior & Organization, with 11 papers having over 100 citations and 2 papers over 1000 citations (two of the three most cited papers since the inception of Economic Quarterly). His research achievements have won the Shanghai Philosophy and Social Sciences Award and have been awarded the “Peak Award” or “Pioneer Award” three times by the Open Research Platform for Digital Economy (out of four selections). He has also published over 90 economic commentaries in mainstream media and authored three monographs, co-authoring several others. He has led over ten projects including the National Social Science Fund Youth Project, Shanghai Philosophy and Social Science Planning Project, and Postdoctoral Science Fund General Project.
In this lecture, Professor Guo Feng elaborated on the fundamental principles of machine learning through three case studies: text sentiment analysis, social media network construction, and the application of machine learning in synthetic control, explaining the important role of machine learning in empirical research in economics and ultimately sharing valuable learning insights and suggestions.
At the beginning of the lecture, Professor Guo Feng started with the question “Why learn machine learning?” He explained that in the era of digital economy, massive traceable big data needs to be simplified with the efficient tool of machine learning. He pointed out that the core of machine learning lies in its predictive capabilities and raised two key issues in machine learning from the perspective of econometrics—overfitting and underfitting. On this basis, Professor Guo detailed the basic principles of machine learning and introduced strategies such as regularization techniques, as well as the basic principles of other machine learning algorithms like ensemble learning and deep learning, focusing on the limitations of least squares in terms of generalization ability.
Next, Professor Guo Feng used three of his research papers as examples to demonstrate the applications of machine learning in different fields: (1) In text sentiment recognition, the predictive capability of machine learning holds not only commercial value but also significant academic importance; (2) Through word embedding models, machine learning can identify semantically similar words and spelling errors, thus comprehensively constructing social media networks; (3) The synthetic control method based on machine learning has more relaxed parameter constraints, optimizing the strict requirements of traditional methods on synthetic coefficients, covariates, and outcome variables.
At the end of the lecture, Professor Guo Feng shared his research experience in the field of machine learning and imparted learning experiences and suggestions to the students present. He emphasized that re-examining traditional problems in the context of the digital economy using machine learning methods is where innovation in research lies. Professor Guo shared his research journey and recent achievements, discussing the potential applications of machine learning in the field of economics. He believes that mastering this tool is more important than the results themselves, as machine learning provides solid support for academic research. Professor Guo offered three practical suggestions: (1) first master Python programming, then delve into machine learning algorithms; (2) learn algorithms one by one, balancing theory and practice; (3) primarily use online resources, supplemented by systematic learning from textbooks. After the lecture, during the interactive session, everyone showed great interest in the application of machine learning in economics. Professor Guo expressed his hope that through this lecture, everyone could understand the essence and applications of machine learning, and what machine learning truly is.
Written by: Seventh Session Xiangzhang Green Seedling Student, PhD Student in Economics – Agricultural, Food and Resource Economics Joint Degree at Michigan State University, Shen Yuxin
Photography: Seventh Session Xiangzhang Green Seedling Student, Undergraduate Student at Macalester College, He Zhijun
Light Up “In View”👇