I-MINER环境下聚类分析算法研究与实现 聚类算法DBSCAN和FCM算法实现.docVIP

I-MINER环境下聚类分析算法研究与实现  摘 要 由于计算机和数据采集技术的进步，数据挖掘获得了非常广泛的应用。数据挖掘帮助用户发现隐藏在大型数据库种的规律和模式，它融合了人工智能、统计、机器学习、模式识别和数据库等多种学科的理论、方法与技术。 数据挖掘模型包括决策树、关联规则、聚类、神经网络、粗糙集、概念格、遗传算法、序列模式、贝叶斯、支持向量机、模糊集和基于案例的推理。其中，聚类是数据挖掘领域的核心技术，被广泛应用于相似搜索、顾客划分、趋势分析、金融投资和信息检索等领域。 I-Miner是一个企业级的数据挖掘工具。在I-Miner的环境下可以方便的建立数据挖掘模型，并且能够嵌入其它数据挖掘模型。对于一个数据挖掘的非专业用户，都能够用I-Miner实现知识的发现。 聚类分析是一种非监督型知识发现的方法，能有效的处理大量的、繁杂的、属性众多的且没有类标志的数据。聚类的目标是在没有任何经验知识的前提下，将数据聚集成不同的簇，使得相同簇中的元素尽可能相似，不同簇中的元素差别尽可能大。通过聚类，人们能够识别密集的和稀疏的区域，因而发现全局的分布模式，以及数据属性之间有趣的相互关系。DBSCAN算法能实现任意形状的数据集的聚类，模糊C均值适合于那些在簇中心周围呈均匀分布的数据集，CABOSFV算法对于高维稀疏数据集（如WEB数据）能很好地聚类。本文在I-Miner中嵌入DBSCAN、CABOSFV和模糊C均值三种聚类分析算法，能够极大地满足用户的需要，建立数据挖掘模型，支持生产决策。 关键词： 数据挖掘；聚类分析；DBSCAN算法；模糊C均值；CABOSFV算法 Abstract As the development of computer and data gathering technology, the data mining is widely used. Data mining which fused the theories, method and technology of many kinds of disciplines such as artificial intelligence, statistics, machine learning, pattern recognition and database help user find the rules and pattern which plant in the large-scale database. Data mining model includes decision tree, connection rule, clustering, neural network, rough collection, concept standard, genetic algorithm, sequence pattern, Bayes, support vector machines, fuzzy set and inference based on case. Clustering is the core technology in data mining domain, and is widely used in the similar search, the customer category, trend analysis, financial investment and information retrieval. I-Miner is an enterprise data mining tool, which might link up with the software which you are using. Building the data mining model based on the I-Miner is simple and inserting other data mining model. Regarding as a non-data mining specialized user, one can use I-Miner to realize the knowledge discovery. The clustering analysis is a non-inspector general knowledge discovery method, which can effectively process massive, numerous and diverse, the attribute numerous, and unlabeled data. The goal of clustering is to g