Stock price forecasting using machine-learning methods

Abstract

The stock market is essential in the economic growth of the nations in which it operates, and stock price prediction is of great significance to investors and government departments, as stocks provide both high reward and high risk. Nowadays, stock price prediction makes extensive use of machine learning algorithms. A large number of machine-learning models are available for predicting stock prices in the existing literature. In this article, the K-Nearest Neighbor (KNN), Random Forest (RF), Long Short-Term Memory (LSTM), and Gate Recurrent Unit (GRU) methods are applied to construct models to make stock forecasting based on Airbnb's historical stock data. The stock data are collected from 10th December 2020 to 19th August 2022. In addition, the accuracy of these four different models is analyzed and compared through Mean Square Error (MSE), Mean Absolute Error (MAE), and Resolvable coefficient (R^2) score metrics. The result shows that the LSTM and GRU models perform better than KNN and RF, with GRU showing the best results.