Guo, Qiulei
(2018)
A Methodology with Distributed Algorithms for Large-Scale Human Mobility Prediction.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
Abstract
In today’s era of big data, huge amounts of spatial-temporal data related to human mobility, e.g., vehicle trajectories, are generated daily from all kinds of city-wide infrastructures. Understanding and accurately predicting such a large amount of spatial-temporal data could benefit many real-world applications, e.g., efficient transportation resource relocation. However, the mix of spatial and temporal patterns among these activities and the scale of the data (in a city level) pose great challenges for accurate predictions under real-time constraints.
To bridge the gap, this dissertation proposes a methodology for the prediction of large-scale human mobility, especially a city level’s vehicle trajectory distribution across the road network. The thesis has several major components: (1) a novel model for the prediction of spatial-temporal activities such as people’s outflow/inflow movements combining the latent and explicit features; (2) different models for the simulation of corresponding flow trajectory distributions in the road network, from which hot road segments and their formation can be predicted and identified in advance; (3) different MapReduce-based distributed algorithms for the simulation and analysis of large-scale trajectory distributions under real-time constraints.
First, our proposed methodology quantifies the latent features of spatial environments and temporal factors through tensor factorization, given existing mobility datasets. We model the relationship between spatial-temporal activities and the latent and other explicit features as a Gaussian process, which can be viewed as a distribution over the possible functions to predict human mobility.
After the prediction of overall inflow/outflow, we further model these movements’ trajectory distributions in the road network, from which the corresponding hot road segments and its possible causes, among other things, can be predicted in advance. For example, based on our prediction, in the next half hour, a high percentage of vehicles that travel from region A/B toward region C/D might pass through the same road segment, which indicates that a possible traffic jam or bottleneck could form there later. This process is computationally intensive and would require efficient algorithms for real-time response because the scale of a city’s road network and the possible number of trajectories that people might choose to take during certain time periods could be very large. Efficient distributed algorithms are proposed and validated.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
24 January 2018 |
| Date Type: |
Publication |
| Defense Date: |
3 November 2017 |
| Approval Date: |
24 January 2018 |
| Submission Date: |
20 January 2018 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
107 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
School of Computing and Information > Information Science |
| Degree: |
PhD - Doctor of Philosophy |
| Thesis Type: |
Doctoral Dissertation |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
spatial-temporal data mining, distributed computing, human mobility prediction |
| Date Deposited: |
24 Jan 2018 16:27 |
| Last Modified: |
24 Jan 2018 16:27 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/33718 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[img]](https://d-scholarship.pitt.edu/style/images/fileicons/application_msword.png)
![[feed]](/images/feed-icon-32x32.png){kind=icon}