Application Research On Data Mining Technology In Small & Medium-Scaled Traffic Accidents

With the development of economy and the improvement of people's living standard, taxies take a large proportion of city passenger transportation. Meantime, a lot of traffic accidents appear in the medium and small cities. Therefore, the traffic management must be reinforced and the traffic accidents must be lessened. As we all know, database has been fully used in traffic management area, which has accumulated a large number of data in this area, especially in the area of coping with the accidents. What's more, a lot of data which are valuable and have potential association exist in the database. It has become an important scientific research task to apply data mining technology to transportation area and use the technology to mine the traffic accidents data at home and abroad. This thesis is to study the clustering analysis and association rule of the data mining technology furtherly, then apply them to traffic accidents analysis system in the medium and small cities.Firstly, in this thesis, the data mining technology and its algorithm are discussed systematically based on the present situation and the trend of development both at domestic and abroad. According to the characteristic of traffic accidents in the medium and small cities, data mining model is constructed optionally. Then the demand analysis of the system and system design are introduced in detail. Emphasis is laid on methods used to process the accident data on the data preprocessing stage.The core of this thesis is how to apply data mining technology to traffic accident analysis system in the medium and small cities. K-means algorithm for clustering analysis and Apriori algorithm for association rule are deeply studied to analyze the merits and demerits in the algorithm. The innovative part of the thesis is that the problems existing in algorithm are analyzed and the original algorithm is improved. The experiment proves that the improved algorithm is better than the original one. The association rule and the improved on clustering analysis algorithm were used to mine the traffic accident data and the results are analyzed. The validity of the system is proved and the purpose of the experiment is reached.