Stock Price Prediction using Linear Regression and LSTM Neural Network

The stock market has a profound influence on the modern society. Therefore, predicting stock prices is always a hot research topic. In this paper, we use linear regression models and LSTM models based on machine learning to predict the stock price of Amazon. In order to let the algorithm more available for individual investors, we only use the historical stock price of the company as data sources. We use MSE, MAE, RMSE, R² to evaluate those models. We find in the case of limited data sources, the simplest linear regression is even better than LSTM models. In LSTM neural network, Single Layer LSTM performs better than Multi-layer LSTM and Bi-directional LSTM. We also find LSTM algorithms that epoch =200 are more suitable in stock prediction than both algorithms that epoch =50 which are faced with underfilling problem and algorithms that epoch =500 which are faced with overfitting problem. This paper applies Bidirectional LSTM and Multi-layer LSTM into stock price prediction, which also provide the possibility of parameter setting improvement.