Research On Visual Analysis Of Travel Characteristics For Taxi Trajectory Dat

With the development of science and technology,intelligence gradually comes into people’s attention,such as smart home,smart car autonomous driving,etc.In order to respond to the call of the country,smart city is also known by everyone,and reasonable planning of urban pattern and traffic is the main problem facing the construction of smart city.How to reasonably plan urban traffic is the basis of urban planning.Traffic problems directly affect the daily travel activities of urban residents.As the most important part of urban traffic,taxis carry GPS devices to collect vehicle trajectory data.How to process and display these data is a hot topic in current traffic research.This thesis focuses on the analysis of GPS trajectory data of Shenzhen taxis,and systematically studies the analysis methods of trajectory data.The research content is mainly carried out from the aspect of travel characteristics,and the research parts are improved,including the following aspects:(1)In trajectory data research,the large data volume often leads to expensive costs in storage,transmission,and query processing.Trajectory compression has become an effective method to alleviate these costs.Therefore,a new model for compressing trajectory data is proposed,and a trajectory route visual analysis method based on a trajectory compression algorithm is introduced.This method integrates synchronous latitude and longitude distance with the Douglas-Peucker algorithm to solve the problem of the original algorithm only considering spatial information and ignoring temporal information.The method reduces the dimensionality of trajectory data by using trajectory compression algorithms,so that trajectory data can be displayed and analyzed on smaller screens.Additionally,this method uses interactive visualization techniques,such as trajectory sliding and scaling,to allow users to better observe trajectory data.Experimental results on a real taxi trajectory dataset show that this method can help users better understand and interpret trajectory data.Compared with the original algorithm,it has significantly improved compression ratio(Cr)and average error(LL)while having minimal spatiotemporal loss,and can effectively reduce the cost of data storage and processing.(2)To address the problems of slow clustering speed and inability to find suitable cluster centers in traditional density-based algorithms for large-scale data clustering,a hot spot region visualization method based on an improved DBSCAN algorithm is proposed,which can segment data and extract clusters in large-scale data.Firstly,the extended clustering is introduced to enhance the accuracy of clustering on the basis of the traditional DBSCAN clustering algorithm.Secondly,the silhouette coefficient is introduced to accurately determine the values of the two parameters EPS and Min PTs suitable for clustering,which significantly improves the time complexity and DBI index of the improved DBSCAN algorithm compared to the original DBSCAN algorithm,according to experimental results.Finally,a hot spot region visualization system is designed to accurately display the hot spot regions in each time period,which provides important decision-making support for further urban layout and transportation planning and improves speed.