Predicting Financial distress with time-series and market sentiment Integration for solving real world problem - capital a berhad

Tan, Zhi Qi (2025) Predicting Financial distress with time-series and market sentiment Integration for solving real world problem - capital a berhad. Final Year Project, UTAR.

Preview

PDF Download (5Mb) | Preview

Abstract

During the COVID-19 pandemic, Capital A Berhad, previously known as AirAsia, encountered substantial financial difficulties. This raised concerns about its financial health and PN17 classification status. Hence, this study examines the company’s financial distress by combining sentiment analysis with financial data using time-series methods. It investigates how market sentiment, drawn from news outlets and customer reviews, influences financial performance and highlights discrepancies between the Altman Z-score and the government’s PN17 classification. Furthermore, the study analyzes the effects of restructuring announcements and market expansion efforts on the company’s financial outcomes. Therefore, three forecasting approaches are compared: time-series analysis of market indices, company-specific financial data, and market sentiment analysis. In this stage, advanced techniques such as LSTM networks for financial data and market indices, alongside BERT model for sentiment analysis, are utilized to construct predictive models. The study follows the CRISP-DM framework, with performance assessed through metrics like mean squared error (MSE) and confusion matrices. This is to evaluate the model's accuracy and robustness. Thus, by outlining the strengths and weaknesses of each approach, this research offers valuable insights to internal auditors and decision-makers at Capital A Berhad, supporting enhanced risk management and financial forecasting practices.

Actions (login required)