Operational Behaviors of GPS-Taxi Driver
Date Issued
2010
Date
2010
Author(s)
Wu, Pei-Ju
Abstract
Taxi drivers and passengers are normally involved in asymmetric information situation. In order to improve this situation, the concept of automatic dispatching taxi was introduced. On the taxi industry, the GPS-based dispatching system is the best way to detach taxi nowadays, but the fleet size of GPS taxi has not reached an economic scale for sustainable operation. One of the main reasons is that the dispatch logic only takes the passenger’s expectation in considering, but leave the taxi drivers’ wanted out of consideration. However, from past researches, it seems lack of the taxi drivers’ expectation. Therefore, this study aims to use GPS-taxi dispatch database to analyze taxi driver operational behaviors by data mining. To increase the breadth of research, this study uses factor analysis to explore key variables among the 12 operational variables while operating conditions, vacancy rate and operating habits are identified as key variable. Then cluster analysis is applied to classify the taxi drivers’ operational behavior into four categories, namely intelligent type, positive type, free type and passive type. To find the drivers’ business behavior of each category, this study uses the details of the original operational behavior variables and the random sampling analysis to analyze time interval between occupied and the other five variables. Finally, suggestions of improving dispatch logic for each type of drivers suggested based on their behaviors. It is expected that taking the taxi drivers and passengers into account simultaneously, the GPS dispatch system shall have the maximum performance. Moreover, it will also attract individual taxi drivers to join the GPS taxi group and then create a better GPS dispatch taxi business environment.
Subjects
Taxi
Driving Behavior
Data Mining
Factor Analysis
Cluster Analysis
Dispatch Logic
File(s)![Thumbnail Image]()
Loading...
Name
ntu-99-R97521506-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):c26aa1d7a0774b698a55f73cecb55411