Prediction Method Using The Multivariate Linear Regression Model Based On The Factor Analysis And Its Applications On The Share Price Forecast

The stock price is a big concern for most citizens in China. Stock price related research is also a hot topic in the communities of financial, economic and system science. Stock markets in China exhibit serial correlation properties which means that historical data contributes to the stock price. Therefore, we can analyze the historical information to forecast the future price. Thus, this article optimizes the traditional multiple linear regression by using factor analysis model to eliminate the model multicollinearity and improve the data fitting. Compared with existing stock prediction methods, the optimized multivariate statistical analysis model in this article needs less work in collecting data, requires less on the data, and predicts better for most stocks. This article takes a certain day’s opening price, bottom price, closing price, highest price, turnover, trading volume and the next day’s opening price of GUANGZHOU PHAR and Western Mining Company as the independent variables to predict the stock closing price of the next day. The developed model in this article is proved to be better by comparing the predicted closing price results before and after eliminating the multicollinearity.