학술논문
Human Mobility: Prediction and Predictability
Document Type
Conference
Author
Source
2024 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops) Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops), 2024 IEEE International Conference on. :445-448 Mar, 2024
Subject
Language
ISSN
2766-8576
Abstract
Predicting human mobility is of significant social and economic benefits, such as for urban planning and infectious disease prevention, e.g., COVID-19. Predictability, namely to what extent a trustworthy prediction can be made from limited data, is key to exploiting prediction for informed decision-making. Current approaches to predictability are usually model-specific along with a relative measurement, leading to varying approximate results and the lack of benchmark assessment criteria. To address this, this study proposes a model-independent method based on permutation entropy to compute an absolute measure of predictability, in particular to derive the maximum level of prediction. Special emphasis is placed on investigating the sensitivity of the predictability methods to changing data loss rates and data lengths. The method has been evaluated using a public dataset with the mobile data of 500,000 individuals. Initial results show a 92%-tighter than before potential predictability and prove the hypothesis of correlation between the minimum amount of data and the level of accuracy of prediction.