題目：Variable Selection in Economic Models: Perspectives from Machine Learning
報(bào) 告 人：于曉華 教授（德國(guó)哥廷根大學(xué)）
主 持 人：劉承芳 教授
報(bào)告時(shí)間：2025年4月1日 星期二 10:00-11:30
報(bào)告地點(diǎn)：北京大學(xué)王克楨樓107會(huì)議室
Xiaohua Yu is Professor and Department Head of the Department of Agricultural Economics and Rural Development at the University of G?ttingen. He obtained his bachelor degree from Renmin University of China in 2001, master degree from Kyoto University in Japan in 2005, and Ph.D. from the Pennsylvania State University in the U.S. in 2009. His research interests cover agricultural economics, behavioral economics and machine learning algorithms. He serves as associate editor for China Economic Review and in editorial boards for a number of international journals, and as scientific advisory committee members for a number of international organizations such as the Forum for Agricultural Research in Africa, and the German Institute of Agricultural Economics in Transition (IAMO).
報(bào)告內(nèi)容
Variable selection is important for estimation, explanation and prediction for economic models. Current empirical researches often select variables in a “cherry-picking” style. Based on the principles of relevance, usefulness, and importance, machine learning provided many scientific feature engineering algorithms, such as filtering method, wrapper method, embedding method, and extraction method. We will review these algorithms with empirical examples.
