基于属性约简理论的风险资产遴选研究

发布时间：2018-03-16 06:25

  本文选题：属性约简　切入点：属性重复度　出处：《河北师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：资本市场是一个收益和风险并存的市场。在资本市场中投资者面临的首要问题是如何有效地规避风险并获取收益。为了规避风险,投资者,特别是机构投资者往往会选择多种风险资产来构造投资组合。目前我国资本市场上的风险资产数以千计,如何将具有投资价值的风险资产遴选出来是构造投资组合并有效规避风险的核心问题。本文以股票为例,应用属性约简理论筛选出能够有效区分风险资产的最小属性集,并以之为依据,利用聚类分析技术对风险资产进行分类,从中选择表现良好的类别包含的股票构造资产池进行投资,并对遴选出的股票的投资业绩进行检验。全文共分五部分,主要内容如下:引言中介绍了选题背景及意义,梳理了国内外有关投资组合模型及属性约简算法的重要文献,提出了研究内容、研究思路、重点难点及创新点等。第二部分首先介绍了属性约简的相关理论,在对以往属性约简算法进行对比分析的基础上,针对属性约简结果易受到离散化效果的影响,提出了基于属性重复度的改进的属性约简算法,该算法以不同样本的属性重复度为离散化依据,以不改变信息系统的不可分辨关系为属性约简依据,弥补了以往属性约简算法在这两方面的缺陷。第三部分选择沪深300指数中2005年前上市且现今仍在沪深300内的股票,详细介绍了改进的属性约简算法的实现过程:首先按行业分类筛选出表现较好的股票,选择反映股票收益及公司财务状况的诸多指标,利用改进的属性约简算法约简属性,得到各行业的最小属性集,并以此为依据对各行业股票进行聚类分析,选取表现良好的类别包含的股票样本构建资产池。第四部分分别利用蒙特卡洛随机模拟法及滚动均值-CVaR模型给出投资组合权重,对改进的属性约简算法遴选出的股票样本的业绩表现进行检验,两种方法均证明本文提出的改进的属性约简算法遴选出的股票样本具有明显的优势。第五部分总结了本文的主要工作,并对未来研究做出展望。
[Abstract]:Capital market is a market in which income and risk coexist. In capital market, the most important problem investors face is how to avoid risk and gain income effectively. In particular, institutional investors tend to choose a variety of risky assets to construct their portfolios. At present, there are thousands of risky assets in China's capital market. How to select venture assets with investment value is the core problem of constructing investment portfolio and effectively avoiding risk. This paper takes stock as an example, applies attribute reduction theory to screen out the minimum attribute set which can effectively distinguish venture assets. On the basis of it, the paper classifies the risky assets by cluster analysis technology, selects the stock structure asset pool which has good performance, and tests the investment performance of the selected stocks. The paper is divided into five parts. The main contents are as follows: the introduction introduces the background and significance of the topic selection, combs the important literatures on portfolio model and attribute reduction algorithm at home and abroad, and puts forward the research contents and research ideas. The second part introduces the related theory of attribute reduction. Based on the comparison and analysis of the previous attribute reduction algorithms, the result of attribute reduction is easily influenced by the effect of discretization. An improved attribute reduction algorithm based on attribute repetition is proposed. The algorithm takes attribute repetition of different samples as the basis of discretization and takes the indiscernibility relation of information system as the basis of attribute reduction. In the third part, we choose the stocks listed in the CSI 300 index before 2005 and are still in the CSI 300 index. The implementation process of the improved attribute reduction algorithm is introduced in detail. Firstly, according to the classification of the industry, the better performance of the stock is selected, and many indexes reflecting the stock returns and the financial situation of the company are selected, and the improved attribute reduction algorithm is used to reduce the attributes. The minimum attribute set of each industry is obtained, and based on this, the stocks in each industry are analyzed by cluster analysis. In part 4th, the portfolio weights are given by Monte Carlo stochastic simulation and rolling mean-CVaR model. The performance of the stock sample selected by the improved attribute reduction algorithm is tested. Both methods prove that the improved attribute reduction algorithm proposed in this paper has obvious advantages in selecting stock samples. Part 5th summarizes the main work of this paper and makes a prospect for future research.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51

【参考文献】

中国期刊全文数据库 前9条

1 李成刚;田益祥;罗聪;徐运保;;基于NBC的定单流与分析师选股收益率比较[J];系统工程;2011年11期

2 何霞;刘卫锋;;基于模糊层次分析法的选股决策[J];长春大学学报;2011年06期

3 马翔;张继福;杨海峰;;基于区分矩阵的启发式属性约简算法[J];计算机应用;2010年08期

4 王天娥;叶德谦;季春兰;;粗糙集属性约简方法在股票预测中的应用研究[J];计算机工程与应用;2009年30期

5 云炜;段禅伦;;基于粗糙集理论的中医方证相关性算法的研究[J];内蒙古大学学报(自然科学版);2007年05期

6 孙士保;秦克云;;基于包含度的决策表属性约简算法的研究[J];计算机工程与应用;2006年03期

7 刘俊,毛道维;基于AHP理论的股票选股模型实证研究[J];电子科技大学学报(社科版);2005年03期

8 林斗志;价值投资在我国股市表现的实证分析[J];财经科学;2004年S1期

9 杨明媚,李华林;主成分分析在证券组合投资中的应用[J];统计与信息论坛;2004年01期

，

本文编号：1618700