Climate change has caused significant warming and increased climate variability worldwide, with Southeast Asia experiencing rapid temperature increases and recurrent extreme events in recent decades. This study employs an unsupervised machine learning approach to classify ASEAN countries based on their climate anomaly profiles between 2004 and 2023. Using monthly temperature anomaly data from the FAOSTAT dataset (relative to the 1951–1980 baseline), two novel indicators were engineered: the Temperature Anomaly Index (TAI), which measures each country’s deviation relative to the ASEAN-wide average, and the Seasonal Temperature Deviation (STD), which quantifies departures from a country’s long-term monthly norm. These features were standardized and analyzed using k-means clustering, with the optimal number of clusters determined through elbow and silhouette evaluation. Results reveal three distinct groups of ASEAN countries: a cluster of nations with moderated anomalies and stable seasonal variability, a cluster characterized by higher relative warming and greater seasonal extremes, and a single-country cluster representing Singapore as an outlier with consistently higher anomalies but lower intra-annual variability. The findings highlight meaningful geographic and climatic distinctions within the region, providing insights into how climate change impacts differ across ASEAN. This work demonstrates the utility of unsupervised learning for regional climate analysis and offers a framework that can guide targeted adaptation strategies, regional cooperation, and future extensions incorporating additional climate variables such as precipitation or extreme event frequencies.
Climate change, Southeast Asia, temperature anomalies, seasonal deviation, clustering, unsupervised learning, k-means, ASEAN