'Modeling Urban Trip Demands in Cloud-Commuting System: An Holistic Approach.'

Thursday, July 06, 2017
12:00 pm

Data Science Colloquium - Menghai Pan, Ph.D. student 


Modeling Urban Trip Demands in Cloud-Commuting System: A Holistic Approach.

Presenter: Menghai Pan (Ph.D. student)

Room: AK013 -  Data Science Innovation Lab


Rapid pace of global urbanization has posed significant challenges to urban

transportation infrastructures. Existing urban transit systems suffer many

well-known shortcomings, where public transits have limits on coverage

areas, and fixed schedules, and private transits are expensive and fail to

timely meet the demand needs. We thus envision a Cloud-Commuting system,

that employs a giant pool of centralized taxis/shuttles to better cope with

the dynamic urban trip demands. To better understand the feasibility of

such a system, in this paper we develop generative models to capture

fundamental demand arrival and service patterns, and introduce a novel

model to estimate the total number of vehicles needed to serve all urban

demands. We conduct experiments using large scale urban taxi trajectory

data from Shenzhen, China, and compare our proposed models with empirical

baselines. We obtained promising results, which shed great lights on future

smart transportation system designs. Besides, we have extended our model to

deploy multiple taxi stations, which takes the picking-up trip and

returning trip of each service into consideration.