数据挖掘方法在股票分析中的应用与研究

发布时间：2018-01-07 23:16

本文关键词：数据挖掘方法在股票分析中的应用与研究　出处：《西南财经大学》2013年硕士论文　论文类型：学位论文

【摘要】：改革开放以来,随着我国经济的快速发展,股市应运而生。我国股市自成立以来,历经了风风雨雨之后,伴随着成长,逐步发展壮大。并且随着人民生活水平的日益提高,人们手头闲置的资金也越来越多,投资需求日益旺盛,投资意识和金融意识也日益增长,投资方式也越来越多样化。股票市场由于其高风险高回报的特性,长期以来,不断吸引人们投入到其中,逐渐成为许多人投资的重要手段之一。越来越多的人将手头的资金投入到股市,以期获得较为可观的回报。然而由于专业知识的缺乏、信息的不对称性等等原因,人们对于股市的投资往往带有盲目性、投机性,很难获得可观的收益。因此,对于股票市场,寻求一套有效的方法,降低人们投资的风险,同时提高人们投资的收益就显得非常重要。在股票市场中,时刻都会诞生大量的数据,上市公司也会定期发布大量的财务数据,如何有效地利用这些数据,减少投资者的投资风险,从而给投资者带来较高的回报便成为了一个非常值得分析研究的问题。上市公司定期发布的财务报告数据具有较大的信息含量,这些信息含量包括各种财务比率指标。综合这些财务指标,能够一定程度上反映公司整体的经营运行状况,有利于投资者判断公司的内在价值,从而有助于投资者更好地判断上市公司股票的投资价值。对于中长期的投资者来说,如何利用这些信息来判断股票的未来投资价值,显得尤为重要。而本文试图通过数据挖掘技术,来研究上市公司公布的财务比率指标和股票投资价值的内在联系,期望挖掘出财务数据中有用的信息,从而对股票的投资价值作出更好的判断。传统的统计模型对于数据有较高的要求,对于数据的假定较多,要求过于严格,实际中的数据往往很难达到这种要求,而数据挖掘技术对于数据的要求相对较低,能够相对较好地处理非正态、非平稳、高噪声的数据。数据挖掘通过结合统计学、机器学习和人工智能等技术对于处理海量数据和高频数据往往能够达到不错的效果。另外数据挖掘还能够对不断获得的新数据进行模型的动态更新,非常适合应用于新环境。数据挖掘是当今国际上统计学、人工智能和数据库研究方面最富活力的新兴领域,从大型数据库中挖掘有效信息的问题已经成为近年来数据分析研究领域中的一个新热点。股票投资风险与机遇并存,如何把握风险,使投资回报最大化是投资者追求的目标。在上市公司公布的财务数据以及股票行情数据库中积累了大量的历史数据,如何充分利用这些历史数据,为投资者提供决策依据,把数据挖掘方法运用于股市投资研究和探索变得很有意义。因此,本文尝试用数据挖掘中的方法来对上市公式财务数据进行分析,目的是发现公司财务数据和股票投资价值的联系,为投资者提供参考。本文基于国内外研究成果,介绍了数据挖掘的相关理论,并且引入数据挖掘的相关方法对上市公司定期公布的财务比率指标和股票价格变化之间的关系进行了研究分析。文章中用到的数据挖掘技术包括决策树分类、神经网络模型以及logistic回归模型三种方法,将三种方法运用于股票价值投资分析中,通过三种方法来研究上市公司公布的财务比率指标与股票投资价值之间的内在联系,并试图寻找哪些财务指标对于上市公司的股价的变化有较大的影响,并且对三种方法取得的结果进行评估和对比分析,比较各种模型进行实证分析时取得的效果,从而更好地判断股票的投资价值。文中建立模型时以上市公司公布的财务指标作为输入变量,为便于不同上市公司的比较,财务指标均选取财务比率指标。并为了综合反映公司的运行状况,从公司盈利能力、偿债能力、发展能力、运营能力以及现金流五个大的方面来选取指标,以更为准确的反映公司的内在价值。此外,以个股赢率作为目标变量建立模型。其中个股赢率为二元变量,当股票一年期的涨跌幅大于大盘指数的涨跌幅时取“1”,否则便取“O”。文章的思路便是以综合反映上市公司运行状况的财务比率指标为输入变量,以个股赢率为目标变量,来研究分析上市公司公布的财务比率指标和上市公司个股赢率是不是存在关系,如果存在关系,哪些财务比率指标对个股赢率的影响较大以及哪种模型预测效果较好,这些都是文章中要研究和解决的问题。
[Abstract]:Since the reform and opening up, with the rapid development of China's economy, the stock market. China's stock market since its inception, after the groundless talk, along with the growth, gradually growing. And with the increasing of people's living standard, people idle funds on hand is increasing, investment demand is increasingly vigorous, consciousness and awareness of financial investment is growing, investment is becoming more and more diversified. The stock market because of its high risk and high return characteristics, long time, continue to attract people into it, has gradually become the important measure to many investment. More and more people would put money into the stock market, in order to obtain a more substantial returns. However due to the lack of professional knowledge, information asymmetry and other reasons, the people to invest in the stock market often with blindness, speculation, it is difficult to obtain benefits. Therefore view, For the stock market, to seek an effective method to reduce the risk of investment in people, it is very important to improve the return on investment. At the same time, people in the stock market, the moment will be the birth of a large number of data, listed companies will regularly publish financial data, how to use these data effectively, decrease the risk of investment thus, to give investors higher returns will become a very worthy of study.
The financial report data of the listed companies has regularly published information content is big, the information content including financial ratios. These financial indicators, to a certain extent reflect the company's overall business operation status, intrinsic value for investors to judge companies, and help investors to better judge the listed company stock investment value. For long-term investors, the future investment value of how to use the information to judge the stock, is particularly important. This paper through the data mining technology, internal relations of financial ratio index and stock investment value of listed companies released, expected to dig out the useful information in the financial data, so as to make better judgment the stock investment value. The traditional statistical model has a higher requirement for the data, the data is assumed to. Pray too strict, actual data is often difficult to meet this requirement, the technology of data mining for data requirements are relatively low, relatively better treatment of non normal, non smooth, high noise data. Data mining by combining statistics, machine learning and artificial intelligence technology for processing massive data and high frequency data tend to be able to achieve good results. Data mining can also dynamically update the model of new data obtained continuously, very suitable for the new environment. Data mining is one of the emerging field of statistics, artificial intelligence and database of the most dynamic, mining useful information from large databases has become a problem data analysis is a new hotspot in the research field in recent years. The stock investment risks and opportunities, how to grasp the risk, to maximize the return on investment is investment The pursuit of the goal. In the financial data released by the listed company and the stock market database has accumulated a lot of historical data, how to make full use of the historical data, to provide decision-making basis for investors, the data mining method is used to study and explore the stock market investment becomes very significant. Therefore, this paper attempts to use the method of data mining to to analyze the financial data of listed formula, the purpose is to find the financial data and stock investment value, provide a reference for investors.
Based on the research results at home and abroad, introduces the related theory of data mining, the relationship between financial ratios and stock price changes and the introduction of related methods of data mining to regularly publish listed companies were analyzed. The data used in the article mining technology including decision tree classification, neural network model and logistic regression model three methods the three methods applied to stock value investment analysis, to study the relationship between financial ratios and stock investment value of listed companies published by the three methods, and try to find out what changes in financial indicators for the stock prices of listed companies have a great effect, and get the results of three methods were analysis of evaluation and comparison, made empirical analysis and comparison of various models of the effect, so as to judge the investment value of stock better in this paper. To set up the model to the listed companies announced financial indicators as input variables, to facilitate comparisons of different companies, financial indicators are selected financial ratios. And in order to reflect the operation status of the company, from the company's profitability, solvency, development ability, operation ability and cash flow five aspects of selection index, to more accurately reflect the intrinsic value of the company. In addition, the stocks win rate as the target variable model. The stocks win rate of two yuan a year variable, when the stock price is greater than the market index rose from "1", otherwise it will take the "O". It is thought to reflect the financial ratios of listed companies operating conditions as input variables, in order to win stock rate as target variable, to study the listed companies announced the financial ratios of listed companies and stocks win rate is not saved In relation, if there is a relationship, which financial ratios index has a greater impact on the winning rate of stocks and what models predict better? These are the problems to be studied and solved in the article.

【学位授予单位】：西南财经大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224

【参考文献】