The Trend Analysis Method of Urban Taxi Order Based on Driving Track Data

摘要

The paper tries to build the analysis framework to explore the implication trend of the complex taxi order. The online taxi has become one of the important means of urban travel with the popularity of the Internet and smart phones. The analysis of online taxi order may contribute to better understand urban traffic trends and people's living habits. The ride-hailing platform can track every order completely through the client, which provides a basis for the analysis of order trend. With the development of big data analysis methods, it is also possible to analyze the trend of urban ride-hailing orders. The research object of this study is the driving tracking data of online taxi orders in Chengdu in October 2016 provided by Didi Chuxing GAIA Initiative. The month covers the China's National Day holiday. And it is the very typical traffic research scenario. This paper analyzed the change trend of urban online taxi order quantity over time, compared the taxi order quantity trends on workday and weekend, and found that workday and weekend order trends about online taxi have structured differently. In addition, k-means algorithm and DBSCAN algorithm were used to analyze the optimal order-waiting location for online taxi drivers, and the comparison between the two methods was made. It is found that DBSCAN algorithm performs better in analyzing such problems. Didi is the largest ride-hailing platform in China, and Chengdu is one of the mega-cities in southwest China. The analysis based on the data of Didi and Chengdu can provide typical research paradigms for the order analysis of urban taxi to some extent.