【作者】 曹阳；

【导师】 陈天滋；

【作者基本信息】 江苏大学 ， 计算机应用， 2006， 硕士

【摘要】 空间数据挖掘与知识发现是指从空间数据库中抽取隐含知识、空间关系和非显式存储在数据库中有意义的特征或模式。该技术在理解空间数据，获取空间与非空间数据间内在的关系方面具有重要意义。近年来，由于地理信息系统被广泛地应用到各个行业中，积累了大量与空间位置相关的空间数据，因此空间数据挖掘的应用研究已经成为当前研究的重要课题。 本文分析了道路交通事故信息管理中的关键问题，即找出交通事故多发地点和造成交通事故的主要原因，这两个问题可以采用空间数据挖掘理论中的空间聚类方法和关联分析方法进行分析。在道路事故的聚类分析问题上，道路事故聚类与一般聚类不同，它是一种障碍空间的聚类问题。为了能够准确地对道路网中的点对象进行聚类，确保同一事故黑点内的事故相关性，本文引入了连通性概念，并设计了空间连通点集的判定算法和事故黑点聚类算法。通过判断各点的连通性来决定是否可以聚类在一起，从而保证了聚类算法在障碍空间的正确性，拓展了聚类分析的应用，实现了障碍空间中的聚类发掘。在事故信息的关联发掘问题上，由于交通事故属性数据是一种多维的数据结构，而一般的关联规则挖掘算法不适合对多维的数据模型进行挖掘。因此，本文对经典的Apriori算法进行了改进，提出了一种针对交通事故属性数据模型分析的多维数据关联算法，增强了算法对于多维空间数据的处理能力。 在以上研究工作的基础上，运用本文提出的算法和ArcGIS Engine组件技术设计实现了交通事故分析原型系统。并采用几组数据进行了分析验证，系统能够对事故数据进行有效的分析，以图形化的方式展现分析结果，大大地提高了对道路事故信息的分析能力。更多还原

【Abstract】 Spatial data mining and knowledge discovery meant to collect the meaningful characteristic or mode among the implicit knowledge, spatial relationship and non-explicit memory in the spatial database, The technology is significant in the areas of spatial data understanding and intrinsic relationship between spatial and non-spatial data capturing. In recent years because of the extensive application of the geographical information system (GIS) in each field and the large accumulation of spatial data related spatial position, the study of how to apply the spatial data mining has already been a significant subject in present research.There are key problems analyzed about the road traffic accident information management in the paper, which is to find out the frequently-occurring places of the traffic accident and main reason to cause the traffic accident, which can be analyzed with the spatial clustering method and association analysis method in the spatial data mining theory. On the clustering analysis of the road accidents, different from general clustering, it was a kind of clustering question of obstacle space. In order to carry on the clustering to the point object in the road network accurately, and guarantee the accident relevance in the same accident black point, the paper introduced connectivity concept and designed the algorithm of spatial connective point sets determined and the algorithm of accident black point clustering. Distinguishing if they could be clustered together by determining the connectivity of each point guaranteed the exactness of the clustering algorithm in obstacle space, expanded the application of clustering analysis, and realized the clustering mining in the obstacle space. On the question of the associated mining in accident information, the attribute data of the traffic accident are a kind of multidimensional datum structure, and the general mining algorithm of association rule was unsuitable to excavate to the multidimensional data model. Therefore, with the improvement of classical Apriori algorithm, there was a multidimensional associated algorithm directed against the datum model analysis of traffic accident attribute proposed, which strengthened the ability to cope with the multidimensional spatial data.On the basis of the above study, with the algorithm pointed in the paper and thetechnical design of ArcGIS Engine, the traffic accident analysis prototype system realized. It analyzed and verified data with several groups, analyzed effectively and showed the analysis result by way of figure, which improved the analytical capacity of the road accident information greatly.更多还原