The Application Of Partial Linear Model In Stock Price Prediction

Securities investment especially the stock market is becoming the important link of theeconomic society along with the rapid development and improvement of economy in China.Stock is the product of economy market. The issuing and trading of stocks are promoted tothe development of economy market. The definition of stock forecast analysis is adetermination for the future development of stock by the scientific method. It is basis on theaccurate statistical data and history、current situation、information of stock market. Stockmarket still exist many uncertain factors from the current situation of stock trading market inchina. At the same time, the stock prediction is an important premise of rational and scienceinvestment. According to the specific situation, the main purpose of this paper is as follows：The first, this paper introduces the basic knowledge related to stock, especiallyintroduces the main financial indicator which influences the stock price. This paper alsointroduces the method of stock price prediction, and analyzes the advantages anddisadvantages of each method on the basis of the above knowledge.The second, this paper describes the application of statistical model in stock prediction.There are including partial linear model and principal component analysis. Part of the firstpartial linear model is outlined, and discusses the least squares estimation of partial linearmodel, finally introduces the principle of principal component analysis.The third, in this paper, through the China Galaxy Securities Neptune stock analysissoftware, obtain the sixty traditional energy sector stocks. The financial indicators and stockprice of the remaining stocks witch for the removal of three ST stock and missing data andabnormal stock as the sample data of regression analysis in December30,2010. Usingcorrelation analysis and other methods to company financial indexes of stock for screening,followed by using the method of principal component analysis to filter out financial indexdata reduction dimensionality, finally we fitting the data of partially linear models based onreduced dimension. According to publish financial index from the September30,2011, weapplicant this statistical model to predict the stock price. The stock price to show the partiallylinear model appears to be better than the linear model.