Data Mining Clustering Algorithms In The Economic Development Of The Industrial Park In Comparison

With the development of computer science, the demand and supply of dataanalysis is growing increasingly, so the traditional data analysis is challenged, whichexpanded rapidly to become an independent field. A meaningful grouping is one ofthe most basic data mining. Cluster analysis is one of typical classification method, itis based on a set of physical or abstract objects, which is grouped different objectsinto multiple categories of similar objects.Multiple indicators are discussed seriously in this essay，which is divided intofive parts:The first part is the introduction, It mainly include: the significance of the topicsin this essay, the dynamics of research and cluster analysis research ideas based onAHP and principal component analysis for dimensionality reduction.The second part is an overview of the cluster theory, which mainly includes theconcept of data mining and the clustering algorithm and exposition of Minimaxlinkage for cluster analysis to compare with various clustering distance linkage.The third part is a description of research ideas and methods, this chapter isbased on statistics theory and combined park actual with the theoretical study of thecluster analysis to divide the park. It is included articles instructions, indicators ofdesign and extraction (dimensionality reduction). This Chapter focuses ondimensionality reduction methods——AHP and principal component analysis.The fourth part is cluster analysis of the park economic development. Thischapter is mainly based on one to three chapters, using principal component analysisand AHP to dimensionality reduction. Comparison with the six kinds distancealgorithms(single Linkage method, complete method Linkage, centroid method,group average Linkage, sum of squares method, Minimax Linkage method) to selectthe best one——Squared deviations （The results are divided into threecategories）.Finally, this article gives the reasons for the differences and improve policy inclination——From Park clustering diagram.The fifth part is summary, that explore the basic content and further work.All in all, selection of clustering algorithm is realistic and comprehensive, whichhas a larger significance.