Prediction of Facebook and GOOG Prices based on Linear Regression and LSTM Regression

Abstract

Stock market analysis is a very difficult task, and stock markets are very complex and constantly changing environments. More and more stock investors are now becoming aware of the prominence of machine learning in the field of stocks and finance, and over the last decade or so machine learning has driven advances in the stock market, such as the ability to use different machine learning methods to predict stock movements in order to arrive at the best decisions and algorithmic trades. The problem that this project wants to investigate is the use of machine learning methods for stock prediction. Two stocks, Facebook and GOOG, were chosen as the datasets for the study. The datasets consisted of stock information from the last decade or so and two machine learning methods, namely long and short term memory and linear regression, were used to make predictions. The results obtained from these two models were analyzing and different results were obtained. The results present the conclusion that the linear regression model is more suitable than the LSTM model for predicting these two groups of stocks. Some error analysis was also carried out and some improvements were given for the two different models.