Analyzing Bangkok city taxi ride: reforming fares for profit sustainability using big data driven model

Abstract

With the trend toward the use of large-scale vehicle probe data, an urban-scale analysis can now provide useful information for taxi drivers and passengers. Unfortunately, traffic congestion has become a critical problem in urban cities. Road traffic congestion reduces productivity in transportation services, and the daily profit earned is consequently reduced. This is opposite to the cost of living, which is increasing rapidly. Therefore, these issues are causing difficulties in all occupations in terms of managing daily expenses, particularly for taxi drivers. The taxi driving is classified as low income compared to other occupations. Such facts are a symbol of economic inefficiency. To this end, this study aims to assist taxi agencies and the government in improving taxi driver profits in Bangkok using large-scale data. To deal with these large-scale data, we propose a big data-driven model. With this model, we first calculate costs using a cost–distance algorithm and trip reconstruction. The data are then modeled to understand distance-based profits with respect to the departure time and traffic conditions. Finally, several cost predictive models using machine learning are evaluated using the ground truth from 50 taxis for a 1-month period. The experiment results show that more frequent trips over a short distance yield higher profits than long-distance trips. Finally, a solution to improve taxi driver profits is determined. We also compare the advantages and disadvantages of a unified solution.