Stock price prediction based on multiple linear regression

Abstract

Stock price prediction plays an important role in finance and economics. In general, a rise and fall in the share price influences the investors’ determinations and spurs the interest of the researchers over the years. The existing forecasting methods make usage of both linear and non-linear algorithms. From the share price fluctuating of NVDA, AMD, and INTC, we adopted the model MLR (multiple linear regression) to forecast the stock trend and find the relatives of the three stocks within the fixed period. Aside from this, correlation analysis was carried out, and several indexes and metrics were applied to evaluate the models. According to the analysis, three models are constructed, and two of them are relatively significant after improving. Overall, these results shed light on guiding further explorations of stock price forecasting based on the state-of-art financial and statistical models in the concept of big data analysis.