《CART算法在新浪微博客户分类中的应用研究》》-毕业论文.doc

济南大学毕业论文 PAGE - PAGE II - 济南大学毕业论文 摘 要 基于当今这个高度信息化的时代，数据挖掘技术及数据仓库的高速发展，通过网络平台交流的用户日趋增加，客户分类就成为了当今社会首要解决的问题。 本论文在数据挖掘的理论、方法及技术上，以决策树为建模主要思想，采用决策树中的基于Gini指数的分类和回归树（CART）算法，把新浪微博客户信息转化为属性-结论式的形式，通过构建树、修剪树、评估模型三步骤，将客户进行分类，从而快速准确地区分目标客户及非目标客户。 在数据挖掘技术模型的基础上，将新浪微博的目标客户模型做部分的调整，最终能够得到最优化模型。该模型有着响应时间较短且精度高的特点，若运用到实际生活中能够大大地提升客户分类的效率，那么无论是企业还是个人都将从中受益良多。 关键词： 决策树；目标客户；CART；新浪微博；数据挖掘 ABSTRACT Our society, now is full of information. Based on the rapid development of the data warehouse and data mining technology, competing in network platform increases day by day. So the customer management becomes one of the most important issues. This paper, based on the theory, technology and methods about data mining and got classification tree for the main ideas of the modeling. Made the customer information of Micro-blog into the properties - the Conclusion form, using the CART algorithm of classification tree which based on the smallest of Gini index. By building the tree, pruning the tree and assessing the tree, the customers are classified. As a result, target and non-target customers are distinguished rapidly and accurately. Based on the data mining, the models of micro-blog with target customers do some adjustments, then we can finally get the optimization model. The combination of data warehouse model, applied to real life can greatly improve efficiency, in other words, the customer or the company will both benefit lots from this. Key words：Decision Tree；Searching Target Customers；CART；Micro-blog；Data Mining 目 录 TOC \o 1-3 \h \z \u HYPERLINK \l _Toc357985839 摘要 PAGEREF _Toc357985839 \h I HYPERLINK \l _Toc357985840 ABSTRACT PAGEREF _Toc357985840 \h II HYPERLINK \l _Toc357985842 一、前言 PAGEREF _Toc357985842 \h 1 HYPERLINK \l _Toc357985843 （一）研究背景 PAGEREF _Toc357985843 \h 1 HYPERLINK \l _Toc357985844 （二）选题目的 PAGEREF _Toc357985844 \h 2 HYPERLINK \l _Toc357985845 二、数据挖掘与客户分类概述 PAGEREF _Toc357985845 \h 3 HYPERLINK \l _Toc357985846 （一）关于数据挖掘 PAGEREF _Toc357985846 \