预测人类移动行为的介入机会类模型研究进展

Research advances in intervening opportunity class models for predicting human mobility

预测地点间人类的移动在人类迁徙、交通预测、疾病传播、商品贸易、社会交往等诸多方面具有重要的意义.介入机会模型是最早从个体目的地选择行为角度建立的预测人类移动的模型,它将起终点之间的介入机会作为影响人类移动的关键因素,启发研究者提出了许多新的介入机会类模型.介入机会类模型在很多学科领域也获得了广泛的应用.本文首先对包括介入机会模型、辐射类模型、人口权重机会类模型、探索类介入机会模型和统一机会模型等在内的介入机会类模型的研究进展进行综述,然后对这些介入机会类模型在空间交互和疾病传播方面的应用进行介绍,最后对该类模型未来的研究方向进行探讨.

Predicting human mobility between locations is of great significance for investigating the population migration,traffic forecasting,epidemic spreading,commodity trade,social interaction and other relevant areas.The intervening opportunity(IO)model is the model established earliest from the perspective of individual choice behavior to predict human mobility.The IO model takes the total number of opportunities between the origin location and the destination as a key factor in determining human mobility,which has inspired researchers to propose many new IO class models.In this paper,we first review the research advances in the IO class models,including the IO model,radiation class models,population-weighted opportunity class models,exploratory IO class models and universal opportunity model.Among them,although the IO model has an important theoretical value,it contains parameters and has low prediction accuracy,so it is rarely used in practice.The radiation class models are built on the basis of the IO model on the assumption that the individual will choose the closest destination whose benefit is higher than the best one available in origin location.The radiation class models can better predict the commuting behavior between locations.The population-weighted opportunity class models are established on the assumption that when seeking a destination,the individual will not only consider the nearest locations with relatively large benefits,but also consider all locations in the range of alternative space.The population-weighted opportunity class models can better predict intracity trips and intercity travels.The exploratory IO class models are built on condition that the destination selected by the individual presents a higher benefit than the benefit of the origin and the benefits of the intervening opportunities.The exploratory IO class models can better predict the social interaction between individuals,intracity trips and intercity travels.The universal opportunity model is developed on the assumption that when an individual selects a destination,she/he will comprehensively compare the benefits between the origin and the destination and their intervening opportunity.The universal opportunity model presents a new universal framework for IO class models and can accurately predict the movements on different spatiotemporal scales.The IO class models have also been widely used in many fields,including predicting trip distribution in transportation science,modeling the purchasing behaviors of consumers in economics,detecting complex network communities in network science,measuring spatial interaction in economic geography and predicting infectious disease transmission in epidemiology.This paper focuses on the applications of IO class models in spatial interaction and epidemic spreading,and finally presents the discussion on the possible future research directions of these models.