A Data-Driven Approach to Flood Risk Assessment and Public Sentiment Analysis

Abstract

Flooding remains as the most frequent and destructive natural disaster impacting Malaysia which causing significant disruptions across social, economic and environmental systems. This study addresses the need to integrate physical risk assessment with public sentiment analysis to strengthen disaster management. Historical flood records from 1967 to 2023 were analyzed together with flood-related news articles, utilizing geographical risk mapping and transformer-based as well as keyword-driven sentiment analysis. The results identified Kelantan and Terengganu as the highest-risk states and revealed dominant emotions of fear and frustration in media coverage, with 74.5% of articles emphasizing rescue operations and only 11.8% focusing on recovery. These findings highlight critical gaps in long-term flood resilience communication and planning. By integrating data-driven flood risk assessment with sentiment insights, the study offers a more comprehensive understanding of flood impacts, supporting more targeted disaster preparedness, communication strategies and policy development in Malaysia.