| The trip of urban residents affects the whole urban life.The analysis and exploration of residents5 trip characteristics is a hot issue in the study of Urban geography.On the one hand,various forms of massive transportation data provide data for comprehensive analysis of trip characteristics.On the other hand,the space-time analysis model provides a tool for revealing the trip characteristics,which can solve the interaction between time,space and attributes effectively.At present,a single type of transportation data is used for the study.However,due to the diversity of trip modes,a single type of data cannot fully reflect the trip characteristics.Most traditional analysis methods are based on time-based time series analysis or distance-based spatial analysis of a single dimension.It is impossible to satisfy the need for trip behavior that has both temporal and spatial characteristics.At the same time,there are few researches on the interactions between different regions,and most of them show a point-to-point model analysis.In view of this,this article constructed a trip space-time data model,a trip hotspot hierarchical structure model and movement patterns between zones algorithm to work on mining trip characteristics in different time nodes,i.e.weekdays,weekends and holidays,based on the multi-source public transportation big data in Shanghai.The main research contents and conclusions are as follows:1.Multi-source public transportation big data preprocessing method based on HadoopThe Hadoop platform can perform distributed parallel computing on massive data efficiently and cost-effectively.The transportation big data used in this study including taxi trajectory data and rail transit smart card data up to 523 GB are suitable for processing and analysis by the Hadoop.Therefore,this study proposed a method of preprocessing subway and taxi trajectory based on Hadoop,and another method for extracting OD dataset,which will lay a foundation for subsequent experiment analysis.The Hadoop platform can perform distributed parallel computing on massive data efficiently and cost-effectively.The transportation big data used in this study including taxi trajectory data and rail transit smart card data up to 523 GB are suitable for processing and analysis by the Hadoop.Therefore,this study proposed a method of preprocessing subway and taxi trajectory based on Hadoop,and another method for extracting OD dataset.2.Constructing a framework for mining trip characteristicsThe characteristics of residents’ trips mainly include three aspects,spatial clustering performance,trip direction distribution and interaction characteristics.Based on these,this study constructed a framework for the processing of trip characteristics.Firstly,using the space-time cube model to visualize the spatial distribution of trip trajectories in three-dimension,and using spatial-temporal hotspot analysis method,time and space are placed together to quantitatively express the clustering degree of high and low values of trip events.Secondly,a hotspot hierarchical structure model was constructed to characterize the hierarchical relationship and the standard deviation ellipse was used to determine the direction of the hotspots distribution,then discuss the spatial structure distribution of the trips.Finally,the movement pattern between zones algorithm was implemented to identify the travel gathering zones and the mode of directional movement between zones,and then construct multiple indicators to extract the interactive characteristics including coverage,dependency and interaction strength.For the hotspot hierarchical structure mode,this study proposed a hotspot detection model based on the trip density field.The study found that the spatial and temporal characteristics of trips can be more fully explored from the above three aspects.3.Mining Residential Mobility characteristicsIn this study,the data of a full month was used to explore spatial characteristics of residents’ trips.It was found that residents exhibited three different trip characteristics on working days,weekends and holidays.At the same time,exploring subways and taxis trajectory can obtain complementary trip characteristics.Subway features are spread over a wide range,and the overall characteristics can be better reflected,while taxi features are concentrated,and more local characteristics can be reflected.(1)Mobility space-time hotspot mode.Subway hotspot patterns can reveal the different commuting time distributions of commuting objects,identify general work areas and residential areas.And hotspot mode of weekends and holidays can identify business circles and transportation hubs.However,the taxi hotspot patterns under different time nodes can identify the four city sub-centers in Shanghai.(2)Hotspot scale relationship and direction differentiation feature.The subway trip hotspots of different grades show obvious pattern grouping characteristics at different time nodes.From the direction of distribution,subway trip hotspots reflect that the general direction of Shanghai’s development is northeast-southwest,and it gradually develops to the west and southwest.While the taxi hotspots have a fixed distribution,which can better reflect the location of the most prosperous areas and transportation hubs in the city.(3)Movement patterns between zones.The workday reflects the commuting behavior of the movement patterns,distributed in the downtown and the Pudong New Area.The weekend patterns are more categories including commuting,sports and entertainment.The significant movement patterns of the holidays are distributed in the peripheral area of Shanghai,which embody the outbound trip.And the patterns include eight single-site zones,indicating that the purpose of holiday trip is more clear.At the same time,the statistics of the indicators of the regional movement model show that the trip zones on the working day are more diverse.And the regional interaction intensity at different time tends to decrease steadily,and the strength of interaction varies from strong to weak,followed by work days>weekends>holidays. |
PDF Full Text Request