基于因子分析的多元线性回归方法及其在股价预测中的应用

发布时间：2018-04-22 13:36

本文选题：多元线性回归 + 多重共线性　；参考：《南京大学》2014年硕士论文

【摘要】：股票价格是中国绝大多数公民关心的问题,也是金融、经济、系统科学等领域研究的热点问题。由于中国股票市场的股票价格时间序列是序列相关的(即历史数据对股票的价格形成起作用),因此,我们可以通过对历史信息进行分析来预测未来的股票价格。基于此,本文即在传统的多元线性回归基础之上,利用因子分析模型进行优化,从而消除模型的多重共线性而达到更好的拟合效果。相较于现有的股票预测方法,该方法数据搜集简便且对于数据的选取无特定要求,预测结果拟合度高,适用于大多数股票。本文以广州药业和西部矿业两只股票的历史价格为实例,以当日开盘价、最高价、最低价、收盘价、成交额、成交量及次日开盘价为自变量,预测该股票的次日收盘价,通过对比消除共线性前后的两个模型对于收盘价的预测结果,验证了利用因子分析模型消除共线性后的回归方程预测效果更好。
[Abstract]:Stock price is a hot issue in the fields of finance, economy, system science and so on. Because the time series of stock price in Chinese stock market is sequence-dependent (that is, historical data play an important role in the formation of stock price), we can predict the stock price in the future by analyzing the historical information. Based on this, based on the traditional multivariate linear regression, the factor analysis model is used to optimize the model so as to eliminate the multiple collinearity of the model and achieve a better fitting effect. Compared with the existing stock forecasting methods, the proposed method is easy to collect and has no specific requirements for data selection. It is suitable for most stocks because of its high fitting degree. This paper takes the historical price of Guangzhou Pharmaceutical Industry and the West Mining Industry as an example, taking the opening price, the highest price, the lowest price, the closing price, the turnover, the volume and the opening price of the next day as independent variables to predict the closing price of the stock the next day. By comparing the prediction results of two models before and after elimination of collinearity to the closing price, it is verified that the regression equation with factor analysis model is better than the other models.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51

【相似文献】