Study On Mining Of Residents Travel Patterns Based On Mobile Location Data

The accelerating of the urbanization process has brought great traffic pressure to large and medium-sized cities.Traffic congestion has become one of the maladies.Studying the regular of travel of urban residents,hot spots of travel and frequent patterns of travel activities provides a way to solve traffic problems.Traditional research methods are generally based on questionnaires and other forms.They have the disadvantages of high financial resources,low efficiency,and small in sample size.As a result,they are unable to fully and accurately identify the activities of residents travel.With the rapid development of modern location technology,mobile location data has become a new and important source of rapid access to real-time location information.Because mobile location data has the advantages of timeliness,low cost,large sample size,etc.,it can fully record residents daily travel activities.The use of a large number of mobile location data to mine the purpose,interests and hobbies,and behavior patterns of residents travel has become a research hotspot in the field of intelligent transportation in recent years.This paper focuses on the spatio-temporal distribution characteristics of Mobile location data,the resident travel pattern is comprehensively studied from the overall and local perspectives.First of all,we have established a variety of travel characteristics indexes from the macro perspective to analyze the travel behavior of residents.According to the relationship between mobile location data,we find that residents have different mobile modes on weekdays and weekends.Then,for the problem that the classic DBSCAN algorithm is difficult to determine the parameters,this paper proposes a parameter adaptive algorithm AC-DBSCAN based on the cutoff distance is proposed,and the parallelization of the algorithm is realized on the Spark platform;on this basis,the weekdays and weekends the mobile location data under different time periods are used to mine,and find the hot spots and spatio-temporal changes regular patterns of residents travel.Finally,aiming at the lack of semantics in the current frequent pattern mining,this paper proposes to take the semantic traffic area as the research object and use the A-PrefixSpan algorithm to mining the semantics of the frequent pattern of residents travel,and obtain the interactive relationship between the residents and the geographic space.Better to discover the knowledge and regular pattern hidden by residents travel.This article uses Python language to implement the relevant algorithms and build a Spark platform composed of seven hosts to process data.Using the three days provided by the mobile company in Guiyang,a total of 54,844,135 mobile location data as a data source to verify the feasibility of this research method.The experimental results show that this paper proposed algorithms are effective and feasible,and the relevant research results can provide more theoretical basis and decision support for urban road traffic planning,traffic operation management,land value assessment,and public facility planning.