Prediction Of Travel Order Volume And Hotspot Clustering Based On Shared Bicycle Data

With the city’s promotion of green and low-carbon travel,shared bicycles have emerged.Shared bicycles have the advantages of low-carbon,environmental protection,and easy borrowing and returning.At the same time,the emergence of shared bicycles has effectively filled the blind spots of public transportation,and the development of shared bicycles can meet the short-distance travel needs of urban residents.Shared bicycles combine the advantages of rail transit with large capacity and stable running time.At the same time,shared bicycles have expanded the scope of influence of rail transit,solved the problem of the "last mile" of travel,and improved the urban transportation system.Although shared bicycles have brought convenience to residents’ daily lives,the research on the travel rules of this new mode of transportation is still insufficient.Shared bicycles are developing rapidly,and their delivery volume and scope of delivery continue to expand.While shared bicycles are convenient for citizens to travel,they also bring new pressures and challenges to urban traffic planning and management and non-motorized vehicle stop location planning.It is urgent to study the rules of shared bicycle travel.Traditional methods such as questionnaires and statistical analysis are no longer sufficient to deal with the current massive data research.Various new technologies brought about by scientific and technological progress have been on the stage of the times.Big data analysis technology and machine learning have been widely used in all walks of life,and they can also analyze and study the time and space data of massive traffic trips in the transportation field.Big data analysis needs the help and support of basic public data.This thesis uses data from Mobike in Chengdu to study the travel patterns of shared bicycle users.The main work of this thesis is as follows:Firstly,perform data preprocessing on the original data,including the coordinate conversion of the original longitude and latitude data of shared bicycle trips,the correction of abnormal GPS trajectory data,and the processing of missing fields.Through preprocessing operations,the target data can more accurately reflect the travel characteristics of shared bicycles and reduce errors in subsequent analysis.Then use the statistical analysis method to analyze the riding characteristics of shared bicycles.According to gender,age,average riding time,and riding distance,user characteristics of Mobike are analyzed.According to the number of orders and calorie consumption,the time characteristics of shared bicycle travel are analyzed.According to the order of shared bicycles riding in different urban areas,the spatial characteristics of shared bicycle travel are analyzed.Next,forecast the demand for shared bicycle travel.The prediction accuracy is improved by adding date-type features and weather features.For the use characteristics of shared bicycles connected by subway stations,the CNN-LSTM prediction algorithm is used for analysis,which improves prediction accuracy.After that,the advantages and disadvantages of the traditional K-means clustering algorithm and the DBSCAN clustering algorithm are compared.This thesis proposes a new clustering algorithm-DB-s Means clustering algorithm.The clustering algorithm proposed in this thesis is used to cluster analysis of Mobike’s cycling orders in the Chengdu area.Furthermore,extract bicycle hot spots.By matching the POI information of the bicycle hotspot location,it provides suggestions for the location of shared bicycle parking.Finally,based on the Spring Boot framework and the Echarts framework,a visualization platform for shared bicycle travel characteristics was developed.The platform can intuitively display the characteristics of shared bicycle travel,which is of reference value for urban planning.