Modeling And Analyzing Of Parking Behavior In Urban Residential Area Based On Data Profile

Parking problem has adverse effects on the sustainable development of urban transportation system.With the depleting of parking space resources,the precision and meticulous management has become an important way to solve the parking problem in urban area.In the realization of precision and meticulous parking management process,the understanding and application of parking behavior feature play a fundamental role.With the continuous development of information technology,traffic big data technology provide a new way of thinking and problem solve method for parking behavior analysis.Therefore,it is an important problem that urban traffic planning and management research needs to solve in order to provide the basis for the development of precise and meticulous parking management measures by using big data thinking and techniques to study the mechanism and the feature of parking behavior.In particular,the background that the rapid development of new urbanization in China and its accompanying motorization level promotes the rapid growth of parking demand,increases the urgency and complexity of solving the parking problem.In view of the above problems,this paper introduces the idea of data profile to analyze and model parking behavior in residential area from the perspective of visualization,labeling and classification and so on.First,taking reference of the user need oriented data profile,the concept of parking behavior profile is provided.The relationship between the features of parking behavior and the operational characteristic of parking lot is qualitatively analyzed by the way of logical route diagram.Based on the requirement of parking behavior profile,the standard flow that preprocessing and normalization of parking lot operation data is established,which provides the basis for further analysis.Secondly,the parking behavior profile is processed based on statistical analysis.Through the statistic and visualization method of parking behavior attribute,the centralized trend of data samples is obtained.The parking behavior feature standard is defined and the typical feature of the different parking behavior is extracted.The parking behavior in residential area is divided into five categories.The duration model of vehicle's paring state is constructed by using the departure time attribute.The distribution features of each type of parking states is extracted to verify the accuracy and rationality of parking behavior classification.Thirdly,the method of machine learning is used to classify the attributes of parking behavior.By using the Gaussian classifier,the probabilities of parking behavior samples belonging to specific parking behavior learning samples are obtained.The grouping of parking behavior is completed by setting the lower limit of probability threshold.Two kinds of methods,which are two-way Gaussian classification and multiple Gaussian classification,are used to obtain four types of parking behavior.The parking behavior is described by parking duration,parking frequency and the entering and departure time.Finally,according to the result of parking behavior profile,we put forward the corresponding precise and meticulous parking management measures in the view of the functional area management in parking lot and the sharing of space and time resource of parking space.We also give full consideration to the practicability under the support of mobile internet technology.