Optimization of Taxi Site Configuration Based on Historical Statistic Demand and Forecasting Trip Amount

• Activity Number: 355
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and Marketing
• Author(s): Wei Zhao*+ and Xinxin Bai and Jinyan Shao and Ming Xie and Wenjun Yin and Jin Dong
• Companies: IBM Research China and IBM Research China and IBM Research China and IBM Research China and IBM Research China and IBM Research China
• Keywords: Optimization of taxi site configuration ; historical statistic demand ; forecasting trip amount ; constrained optimization ; penalty function

Abstract

A novel approach to optimize the taxi site configuration based on historical statistic demand and forecasting trip amount is presented. Considering the limitation of resources, we combine the optimization for configuration of single taxi site with global planning of site configuration in a whole city. The problem for configuration of single taxi site is modeled as constrained optimization problem, which can be solved by combination of penalty function and one-dimensional search methods based on the historical statistic data of its net taxi demand. The global planning of site configuration in a whole city is conducted by imputation method based on two types of data for each site, including surrounding traffic information and forecasting data of surrounding citizens' trip amount by taxi. Finally, an example with results is given to show the effectiveness of our approach.