Font Size: a A A

Improved Clustering Algorithm And Its Application In The ATM Location

Posted on:2015-06-01Degree:MasterType:Thesis
Country:ChinaCandidate:Z H YangFull Text:PDF
GTID:2298330431493055Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the rapid development of the Internet age, the amount of data stored in thedatabase is growing exponentially. In order to find useful knowledge from theselarge amounts data, the technology of data mining applications was born. In thispaper, basic theoretical research mainly used for data mining tasks cluster analysiswas studied. From Partition-based, hierarchical-based, density-based, three differentaspects of cluster analysis algorithm analysis, research different clustering analysisalgorithm running processes, advantages, disadvantages and scope of use. And tocluster analysis commonly used in K-MEANS algorithm has been improved,proposed a new K-MEANS algorithm: using the square error and the lowering speedto automatically determine the optimal value of the parameter k, overcome thetraditional values of the algorithm parameters k difficult issues identified; whenchoose the initial means point, according to the initial mean point distance thresholdto make the initial mean point as discrete, overcome the traditional algorithmiterations during the run fluctuations and easy to fall into local optima. Same time asthe K-MEANS algorithm processing result data by outlier greater impact, in order toobtain a better quality of clustering results, this paper presents a density-based outlierdetection method. Central idea of this method is to determine the neighborhood ofthe data object number of other data objects to determine whether the data object isan outlier.In applied research extension, this paper uses the improved K-MEANSalgorithm applied to the ATM location system. ATM is important financialequipment for bank, its role is very important for the bank and the customer,therefore the research of ATM location practical significance. First, according to therequirements of ATM location, studied the classic commercial facility location modeltheory. Then combination of factors that the actual effect ATM location, ATMlocation evaluation model was constructed. Finally, the improved K-MEANSalgorithm with ATM location evaluation model for solving the location problem,forming the ATM location system.The improved K-MEANS algorithm and ATM location systems in this paperare programming with C#object-oriented coding languages, and running well.
Keywords/Search Tags:Cluster Analysis, K-MEANS Algorithm Improvements, ATM Location System
PDF Full Text Request
Related items