股票信息处理分析系统研究与实现

发布时间：2018-04-29 13:12

本文选题：股票 + 时间序列　；参考：《浙江工业大学》2012年硕士论文

【摘要】：股票的股价序列是一个典型的时间序列，利用时间序列技术对股价序列进行研究分析有一定的理论意义和实用价值。通过研究时间序列相关理论和方法，将其应用于股价序列，可发现股价时间序列的内在变化规律，从而对其进行趋势预测分析。研究利用股价时间序列建模技术，并为投资者提供一个能自动化、智能化分析股市信息的工具，是本文的研究目标。本文的主要研究内容包含了以下几个方面： (1)设计和实现了一个股票信息处理分析系统，能实现用户管理、技术分析、条件选股等常规股票分析功能。 (2)提出了一种改进的适用于股票股价序列的拟合算法，，该算法的思想是采用斜率法和三角中线法相结合的办法来寻找股票的股价关键趋势点作为分段点，进而对序列进行分段线性拟合，最后在行情处理系统中对算法进行实现。在实证研究中，通过与几种常见拟合算法的比较发现，该改进的算法在数据压缩和对股票趋势的提取这两个方面具有更好的效果；最后将此改进的算法融入到股票信息处理分析系统中，具有操作方便，能迅速获取股票关键信息等优点。 (3)针对如何选择输入特征向量能使支持向量机的预测效果更加精准这一问题，本文提出了两种改进的算法：一是运用关键点查找算法来对原始的股票信息进行特征选择，选择股价序列中能代表股票整体走势的序列作为SVM的输入特征向量；二是通过决策树的信息增益法来判定股票输入特征的重要程度，再根据信息增益值来对特征进行加权计算后作为SVM的输入特征向量；最后将此两种改进的算法融入到股票信息处理分析系统中，能在很大程度上提高了预测结果的精度。总之，本文在实现了股票分析的基本功能之上，又结合上述研究的时间序列算法，侧重实现了对于股价序列的基于关键点的SVM预测功能和基于决策树加权特征选择的SVM预测功能，使得系统具有其他股票分析系统所没有的功能。
[Abstract]:The stock price sequence is a typical time series. It has some theoretical and practical value to study the stock price sequence by using time series technology. By studying the theory and method of time series related to the stock price sequence, we can find the internal change law of the time sequence of the stock price, so as to carry on the trend of the trend. The research aim of this paper is to use the time series modeling technology of stock price and provide an automatic and intelligent tool for investors to analyze the stock market information. The main research contents of this paper include the following aspects:
(1) designed and implemented a stock information processing and analysis system, which can realize user stock management, technical analysis, conditional stock selection and other conventional stock analysis functions.
(2) an improved fitting algorithm suitable for stock stock price sequence is proposed. The idea of this algorithm is to use the method of slope and triangular midline to find the key point of stock price as a piecewise point, and then piecewise linear fitting to the sequence. Finally, the algorithm is realized in the market processing system. In the study, by comparing with several common fitting algorithms, it is found that the improved algorithm has two advantages in data compression and the extraction of stock trend. Finally, the improved algorithm is integrated into the stock information processing and analysis system, which has the advantages of convenient operation and quick acquisition of the key information of stock.
(3) in order to select the input feature vector to make the support vector machine more accurate, this paper proposes two improved algorithms: first, using the key point search algorithm to select the original stock information, choose the sequence of the overall trend of the stock in the stock price sequence as the input feature of the SVM The two is to determine the importance of the input characteristics of the stock by the information gain method of the decision tree, and then weigh the features according to the gain value of the information as the input feature vector of the SVM. Finally, the two improved algorithms are integrated into the stock information processing and analysis system, and the prediction results can be greatly improved. Precision.
In conclusion, this paper has realized the basic function of stock analysis, and combined with the time series algorithm of the above research, it lays particular emphasis on realizing the SVM forecasting function based on the key point of the stock price sequence and the SVM prediction function based on the decision tree weighted feature selection, which makes the system have its function that his stock analysis system has not.

【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP311.52;F830.91

【参考文献】