Stock Market Analysis and Prediction Using ARIMA, Facebook Prophet, and Stacked Long Short-Term Memory Recurrent Neural Network

Abstract

Stock analysis involves comparing a company's current financial statement to its financial statements in previous years to give an investor a sense of whether the company is growing, stable, or deteriorating. Stock market analysis helps in getting insights into a company's stock and to make better decisions in buying or selling shares in the stock market. This chapter proposes a method to analyze and predict stock market prices based on historical data of 4 MNCs namely, Amazon, Apple, Google, and Microsoft. The prediction is implemented using three models; namely, ARIMA model, Facebook's Prophet model, and lastly a self-constructed, stacked LSTM model. The results of the three models are compared and analyzed. Mean absolute error is used to analyze the performance of the models on real-time test data. The minimum loss achieved by Facebook Prophet Model is 2.445, by ARIMA Model is 10.782, and the Stacked LSTM Model achieved a minimum loss of 6.552.