Research On City Traffic Distribution And Taxi Travel Route Recommendation Algorithm

Nowadays, more and more people use taxis as their daily transport, and taxis have become an integral part of city traffic. Taxis travelling in a city reflects the distribution of traffic to some extent. However, vacant taxis travelling on the road causes some negative impact on the driver’s profit, city traffic load and environmental pollution. Therefore, improving the utilization of taxis, and reducing the travel time and travel distance of vacant taxis on the road, are very pressing issues.The focus of this paper is to propose a route recommendation algorithm based on the distribution of real-time traffic, which is expected to recommend routes where drivers are likely to find passengers within as little time as possible. After reading a large amount of references, the background and purpose of taxi travel route recommendation is studied, the development and characteristic of recommendation system is expounded, the solution of optimal path planning problem in traffic network is analyzed, and the clustering algorithm in data mining is studied, all of this provide a theoretical foundation for research in this paper. In this article, two strategies of searching the best route are compared, namely, the shortest distance and the shortest time, coming to the conclusion that one important factor that affects the route recommendation is the real-time traffic conditions. In this paper, the GPS data of taxis in Nanjing is extracted and processed to find out the locations where passengers get on taxis in the experimental area, then K-Means clustering algorithm is used to obtain the recommended points according to geographical location, and the method of calculating possibility of finding passengers at these recommended points can be put forward. City traffic distribution can be obtained through the analysis of historical GPS data, and then the average speed of every road in each period can be calculated. In this paper, a PVC (potential vacant cost) equation is proposed to calculate the potential vacant cost of each recommended route, and a most cost-effective route algorithm is designed recommending to vacant taxi drivers. The real GPS data is used to validate the effectiveness of the algorithm, and the performance of the algorithm is analyzed through comparing with other algorithms, then the influence of two constant parameters on the performance of algorithm is studied. At last, the research content is summarized, and future research direction was prospected.