LSTM Based Model For Apple Inc Stock Price Forecasting

Abstract

The prediction of stock price is a popular and difficult topic that attracted and confused many investors over a long period of time. Because of the complex transaction market, there are a lot of risks when we do transactions. Until now, there are two schools about the stock market forecasting: fundamental analysis and technical analysis. The topic of this paper is to use the Recurrent Neural Networks to predict the stock price of Apple Inc in the future. In addition, the important unit of our RNN is Long Short-term Memory (LSTM), which introduces the memory cell, replacing traditional artificial neurons in the hidden layer of the network. Our Networks are able to associate memories and input remote in time, which could grasp the structure of data dynamically over time with high prediction capacity. To visualize our results, we draw three figures. We evaluated our model's performance on the dataset provided by the kaggle competition. The results of the experiment show that our method achieves a good performance compared with other machine learning methods. The RMSE of our model is 0.66 and 0.39 smaller than ridge regression and the neural network model respectively.