基于密度和层次的快速聚类算法在数据挖掘中的设计及实现

本论文在对各种算法深入分析的基础上,尤其在对基于密度的聚类算法he基于层次的聚类算法深入研究的基础上,提出了一种全新的基于密度和层次的快速聚类算法。该算法保持了基于密度聚类算法发现任意形状簇的优点,而且具有近似线性的时间复杂性,因此该算法适合对大规模数据的挖掘。理论分析和实验结果也证明了基于密度和层次的聚类算法具有处理任意形状簇的聚类、对噪音数据不敏感的特点,并且其执行效率明显高于传统的DBSCAN算法。

The Design and Realization of Fast Clustering Algorithm Based on the Density and Level in Data Mining

This paper on the basis of deep analysis in all kinds algorithm, especially based on density clustering algorithm and on the level of clustering algorithm deep research. Proposes a new fast clustering algorithm based on density and level. The algorithm based on density keep clustering algorithm find the advantages of arbitrary shape clusters, and with approximate linear time complexity, so the algorithm suitable for large-scale data mining. Theoretical analysis and experimental results also proved based on density and hierarchical clustering algorithm is deal with arbitrary shape clusters clustering, is not sensitive to noise data, and the characteristic of its efficiency is obviously higher than that of the traditional DBSCAN algorithm.