模糊C-均值聚类在股票投资中的应用

发布时间：2018-02-23 07:54

  本文关键词： 聚类分析 模糊C-均值聚类 β-KFCM 粗糙模糊C-均值聚类　出处：《东北石油大学》2017年硕士论文　论文类型：学位论文

【摘要】：聚类分析是重要数据挖掘算法之一。近年来,随着数据库技术的飞速发展和广泛应用,人们的生产活动产生大量的多元化数据。面对大规模的数据,数据挖掘的一项重要任务就是将其进行合理的归类。聚类分析就是一种有效的将数据合理归类的方法,它可以从大量数据中发现隐含在其中的数据结构。随着我国资本市场的发展,中国股市已经具有一定规模,上市公司数量也有了很大提升。在股市发展初期,股民的投资具有较强的投机性,基本面和技术面的分析已经明显失去了有效性,投资无法获得收益甚至严重亏损,运用会计指标进行理智的选择股票就需要对股票自身的品质进行评价,放眼全球目前能够在资本市场占有一席之地的只有量化投资。本文系统的介绍了模糊聚类方法的国内外研究情况、研究的目的及意义,并且将理论研究与实证研究相结合,分层次介绍数据结构理论、数据样本间相似属性和聚类准则函数。研究了的模糊聚类理论,以模糊聚类分析算法及实现过程为主线;首先研究模糊C-均值理论中数据集C划分方法和模糊C-均值分类法;模糊C-均值聚类算法理论中对模糊核聚类算法和高斯核模糊C-均值目标函数进行求解;结合投资组合理论将投资组合中单个资产对组合贡献率考虑其中同时解释了β的投资意义,提出加权模糊核聚类算法β-KFCM模糊分类法;最后研究了粗糙模糊C-均值基本理论,并简化了粗糙模糊C-均值聚类算法。本文实证部分,选取2015年第四季度和2016年第三季度流动比率、现金比率、营业收入同比增长率、总资产报酬率、销售净利率和资产负债率六个财务指标对A股市场不同板块50支股票进行实证分析,结合股票收益率将50支股票分为三类投资等级,其中Ⅰ类为建议的优质投资品种,Ⅱ类和Ⅲ类次之。
[Abstract]:Clustering analysis is one of the important data mining algorithms. In recent years, with the rapid development and wide application of database technology, people's production activities produce a large number of diversified data. One of the important tasks of data mining is to classify the data reasonably, and clustering analysis is an effective method to classify the data reasonably. With the development of China's capital market, the Chinese stock market has already had a certain scale, and the number of listed companies has also greatly increased. In the early stage of the stock market development, The investors' investment is highly speculative, the analysis of fundamental and technical aspects has obviously lost its effectiveness, and the investment has not been able to obtain income or even a serious loss. The use of accounting indicators for rational selection of stocks requires evaluation of the quality of the stock itself. At present, there is only quantitative investment in the capital market. This paper systematically introduces the domestic and foreign research situation of fuzzy clustering method, the purpose and significance of the research, and combines the theoretical research with empirical research. This paper introduces the theory of data structure, similar attributes and clustering criterion functions among data samples. The fuzzy clustering theory is studied, and the main line is fuzzy clustering analysis algorithm and its implementation process. Firstly, the methods of data set C partition and fuzzy Cmean classification in fuzzy Cmean theory are studied, and the fuzzy kernel clustering algorithm and Gao Si kernel fuzzy Cmean objective function are solved in fuzzy Cmean clustering algorithm theory. Combined with portfolio theory, the contribution rate of single asset to portfolio in portfolio is considered, and the significance of 尾 investment is explained at the same time, a weighted fuzzy kernel clustering algorithm 尾 -KFCM fuzzy classification is proposed, and the basic theory of rough fuzzy C- mean is studied. In the empirical part of this paper, we choose the liquidity ratio, cash ratio, annual growth rate of operating income, total return rate of assets in in the fourth quarter of 2015 and in the third quarter of 2016. The six financial indexes of net interest rate of sale and asset-liability ratio are used to analyze 50 stocks in different sectors of A-share market. The 50 stocks are divided into three types according to the stock return rate. Class 鈪，

本文编号：1526505