Urban Traffic Environment Partition Based On Supervised Classification

Urban traffic environment is one of the main research questions in sustainable development of urban transport system. Urban traffic environment has been researched in the fields of urban transport systems planning, traffic management, traffic control and other fields, and become an important consideration in urban transport system planning, traffic management and control measures.Traffic environment partition will divide the city into different areas of the environment, then optimal and govern it.Supervised classification methods and unsupervised classification methods are used in urban transportation partition, and the situation of urban transportation partition was further researched.The specific content of the paper is divided into five parts, as follows:Chapter 1:Introduction, having done a brief overview of the content of this article, mainly introduces the topic of research background, significance, the research status and so on Chapter 2:Traditional Supervised and unsupervised methods and theories, this chapter describes the supervised classification and unsupervised classification methods and their advantages and disadvantages of the algorithm was introduced. Also introduce the image preprocessing, the principal component analysis and support vector machines the basic theory. Chapter 3:Environmental partition algorithms of urban transportation, urban transportation district in this chapter the algorithm will do a more detailed description, and proposed the establishment of non-supervised classification methods in supervised classification method. Chapter 4: Empirical study, this chapter will use Beijing to analyze the algorithm classification accuracy. Chapter 5:Summary and outlook, this chapter outlines the main contents of this article and innovation, and pointed out the shortcomings and the future development direction.Innovation of this paper are as follows:(1) Added to the traditional method of noise control variables, the data through its potential to control the noise, while the distance function introduced in the process of setting the distance weighted streets, and the selection of weights were discussed.(2) According to the characteristics of urban situation and reality, made more in line with the index system of urban traffic environment, and according to the different situations of missing data, using different methods of data interpolation.(3) According to the special nature of this classification, support vector machine multi-classification methods:binary tree classification and taxonomy-one are combined to a new algorithm.