Predicting households’ residential mobility trajectories with geographically localized interpretable model-agnostic explanation (GLIME)

Abstract

AbstractHuman mobility analytics using artificial intelligence (AI) has gained significant attention with advancements in computational power and the availability of high-resolution spatial data. However, the application of deep learning in social sciences and human geography remains limited, primarily due to concerns with model explainability. In this study, we employ an explainable GeoAI approach called geographically localized interpretable model-agnostic explanation (GLIME) to explore human mobility patterns over large spatial and temporal extents. Specifically, we develop a two-layered long short-term memory (LSTM) model capable of predicting individual-level residential mobility patterns across the United States from 2012 to 2019. We leverage GLIME to provide geographical perspectives and interpret deep neural networks at the state level. The results reveal that GLIME enables spatially explicit interpretations of local impacts attributed to different variables. Our findings underscore the significance of considering path dependency in residential mobility dynamics. While the prediction of complex human spatial decision-making processes still presents challenges, this research demonstrates the utility of deep neural networks and explainable GeoAI to support human dynamics understanding. It sets the stage for further finely tuned investigations in the future, promising deep insights into intricate mobility phenomena.Keywords: Explainable GeoAImodel-agnostic explanationlong short-term memory (LSTM)trajectory predictionresidential mobility AcknowledgementsAny opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.Disclosure statementNo potential conflict of interest was reported by the author(s).Data and codes availability statementThe data, codes, and instructions that support the findings of this study are available on figshare at https://doi.org/10.6084/m9.figshare.21543549.v1Notes1 We have 75 variables in total as categorical variables including state and housing type are input as dummy variables into the models.Additional informationFundingThis research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (Grant No. RS-2022-00165821) and the Faculty of Social Science at Western University. This work was also supported in part by the National Science Foundation under Grant No. 2031407.Notes on contributorsChanwoo JinChanwoo Jin is an assistant professor in the Department of Humanities and Social Sciences at Northwest Missouri State University. He holds a PhD in Geography at the University of California, Santa Barbara/San Diego State University (Joint Doctoral Program). His main research interests include big spatiotemporal data analysis, Geospatial Artificial Intelligence (GeoAI), human mobility and urban dynamics.Sohyun ParkSohyun Park is an assistant professor of Computational and Data Sciences at George Mason University Korea. Her research focuses on the interactions of the moves of people and goods with the local environment. She is also interested in exploring geospatial data through quantitative methods.Hui Jeong HaHui Jeong Ha is a PhD student in the Department of Geography & Environment at Western University. She specializes in the spatiotemporal analysis of human movement and the study of urban and neighborhood changes. Her work seeks to create novel methodologies for geographic knowledge discovery and develop open-source software tools that support human movement research.Jinhyung LeeJinhyung Lee is an assistant professor in the Department of Geography & Environment at Western University. His research interests focus on using GIScience, spatial analysis, and time geography approaches to study urban transportation. Specifically, he aims to develop novel analysis and modeling techniques to support the study of human mobility and accessibility in space and time.Junghwan KimJunghwan Kim is an assistant professor at the Department of Geography at Virginia Tech. His research interests include human mobility (e.g., travel behavior and accessibility), environmental health, the application of geospatial data science methods, and geospatial data privacy/ethics.Johan HutchenreutherJohn Hutchenreuther is a PhD Student in the Department of Geography and Environment at Western University. His research interests are using GIScience, spatial analysis, and time series data to explore neighborhood dynamics and the urban form. Currently, he is focusing on the impacts of transportation on the evolution of neighborhoods.Atsushi NaraAtsushi Nara is an associate professor in the Department of Geography and the Associate Director of the Center for Human Dynamics in the Mobile Age at San Diego State University. He holds a PhD in Geography at Arizona State University. His main research interests include geocomputatio, spatiotemporal data analysis and modeling, human dynamics and movement behaviors, and complex adaptive systems.