Advances and Review of Machine Learning Applications in Urban Studies from 2005 to 2020

Abstract

Machine learning, as a new method for data mining and problem prediction, has been widely used in various fields of urban studies in recent years, which requires a periodical summary of relevant literature. Start with data types, selection and preprocessing, this paper introduces the characteristics and applicability of various machine learning algorithms, and analyzes the cross-fields, hot spots, frontiers and trends of machine learning and urban studies from 2005 to 2020 by using Citespace. Second, focusing on the application of supervised machine learning algorithms from relevant literature in the past five years, a review is made from four main aspects including urban traffic, urban ecology, physical geography, human geography, and the tentative explorations of unsupervised learning, semi-supervised learning and reinforcement learning method in urban studies are unscrambled as well. Finally, the advantages of machine learning methods are summarized, and it􀆳s proposed that the application potential of various machine learning methods in multiple fields and perspectives of urban research should be explored in the future, and the cutting-edge trend of efficient combination of intelligent technology and methods with urban research should be grasped.