Analysis And Exploration Of Passenger Travel Patterns Based On Bus Intelligent Card Data

Currently,with the acceleration of national urbanization,the continuous increase of urban population leads to that the travel needs of citizens keep increasing.However,the development of public transportation in some cities cannot meet the rapidly growing travel needs of citizens.It is a common scene to see urban traffic jams and congestion.Public transportation is the preferred vehicles for citizens.The development of public transportation is an effective measure to solve the problem of urban traffic congestion.In the era of big data,it is meaningful for public transportation mangers to design a system to analyze travel patterns and predict passenger flows.Such a system enables public transportation managers to obtain passenger travel patterns and real-time passenger flow conditions,which can help them adjust their operating and management strategies in time.In addition,such system can also improve the efficiency of urban public transport and get closer to the needs of passenger travel in order to reduce the urban traffic congestion.This paper aims to depict passenger travel patterns better and predict the change of passenger flow in the public transportation.Based on the data of the smart IC card,on the one hand,we qualitatively analyze the collective travelling characteristics of passengers and the community features of the travel network from the spatiotemporal perspective.Then the Pearson correlation coefficient is applied to quantify the dynamic changes of passenger travel modes.On the other hand,on the basis of the time series data mining and the regression forecasting,a combined forecasting model is established to predict the passenger flow of each station.The main contributions of this paper are as follows:(1)In order to reveal the spatiotemporal characteristics of passenger travels,the Pearson correlation coefficient is used to quantify the dynamic changes of passenger travel patterns in the spatiotemporal dimension.Meanwhile,the community structures of the passenger travel networks,which are constructed at different periods,reveal the clustering characteristics of travel routes in a city.First,this paper implements a qualitative analysis of passenger travel patterns from the spatiotemporal perspective.The results show that there exist the morning and evening peak aggregation of passenger travel.The distinct patterns of such aggregation can be found in different stations.Then,the Pearson correlation coefficient is used to quantify the changes in daily passenger travel patterns in a week or those between different stations.It can be seen that the passenger travel patterns between working days are similar,but the differences are obvious between working days and weekend.Moreover,the passenger travel aggregation patterns of adjacent stations have a higher similarity.Finally,the Louvain algorithm is utilized to detect community structures in the passenger travel network.Through analyzing the community characteristics of the travel network at different periods,we find that the community structure of the travel network constantly changes with time.Despite the changes,some stations still reside in the same community.(2)To help predict the passenger flow of subway stations,this paper proposes a combined prediction model ARIMA_LASSO based on time series and linear regression.The construction of this combined model can improve the accuracy of passenger flow prediction.In such combined one,the weight of single model gains optimization.Based on the IC card data of Chongqing Metro,this paper analyzes the prediction results of two single models included in ARIMA_LASSO model firstly,and then the optimal prediction parameters are selected from the single model as the parameters of the combined prediction model.The result shows that the prediction accuracy of the combined prediction model is higher than that of the single prediction model.Also,this paper analyzes the prediction results of the ARIMA_LASSO model at different time granularities.The ARIMA_LASSO model is utilized to predict the passenger flow in the next 15 minutes,30 minutes,and 1 hour,and the result shows that the ARIMA_LASSO model has better prediction accuracy at slightly longer intervals.Thus,this paper takes into consideration that passenger flow forecasting at slightly longer intervals is greatly meaningful for rail transit management departments,which means that more time can be acquired to formulate reasonable operating strategies.Finally,to further verify the prediction accuracy of the ARIMA_LASSO model,the prediction results of recently used models are compared with the model proposed in this paper,and it can be clearly seen that the prediction accuracy of the ARIMA_LASSO model is better than other related models mentioned in this article.By predicting the inbound passenger flow of station,the public traffic manager can obtain the passenger flow status of the station in advance,which can help allocate staff and train resources more reasonably.In summary,based on the IC card data of public transportation,this paper analyzes the characteristics,modes and dynamic changes of passenger travel,and forecasts the passenger flow of station.These analytical results can allow public transport managers to better understand the passenger travel patterns and their dynamic changes.Furthermore,this provides a scientific basis for them to adjust the scheduling strategy and management strategy in time.It can contribute a lot to the development of public transportation.