Stock Market Prediction Using Time Series Analysis: a Case Study of Apple Inc.

Ng, Xin Ying (2025) Stock Market Prediction Using Time Series Analysis: a Case Study of Apple Inc. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text RMM_Ng Xin Ying_Full Text.pdf Restricted to Registered users only Download (3MB)

Abstract

Forecasting stock prices is vital for effective financial decision-making in volatile markets. This study presents a comprehensive framework to predict the daily adjusted closing price of Apple Inc. using both traditional statistical methods and modern machine learning techniques. Data from 2 January 2018 to 31 December 2024 was reindexed to a business-day calendar and enhanced through feature engineering, notably by calculating the high-low range (HL_range) to capture intraday volatility. The modelling approaches employed include variations of autoregressive integrated moving average models—specifically ARIMAX (autoregressive integrated moving average with exogenous variables) and SARIMAX (seasonal ARIMA with exogenous variables)—as well as the Prophet forecasting model, which automatically detects trends and seasonal patterns with minimal parameter tuning. In addition, a long short-term memory neural network (LSTM) was implemented to model non-linear relationships, and several hybrid combinations of these methods were tested to determine the optimal model configuration. Comprehensive data cleaning and exploratory analysis ensured the dataset’s reliability, and an interactive prototype was developed to allow users to select custom date ranges, view forecast visualisations, and export results. Evaluation using error metrics such as root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), directional accuracy, the percentage of forecasts within ±2% error, and the coefficient of determination (R²) demonstrated that the framework delivers accurate forecasts over a one-year period, offering a practical tool for investment planning, portfolio management, and risk mitigation. Keywords: Time Series Analysis, Stock Price Prediction, ARIMAX, SARIMAX, Prophet, LSTM, Hybrid Model, Apple Inc., RMSE, MAE, MAPE, R², ±2%, Directional Accuracy