<![CDATA[This research implements data mining techniques to analyze rainfall patterns through two different approaches: clustering and regression. The first dataset contains monthly rainfall data from East Kalimantan in 2005, analyzed using three clustering algorithms—K-Means, Agglomerative, and MeanShift to identify seasonal patterns. The second dataset uses multiregional rainfall data since 1979, analyzed using three regression algorithms—Linear Regression, Random Forest, and Support Vector Regression (SVR)—to predict June rainfall based on data from the previous five months. Evaluation results show that K-Means and Agglomerative produce the same clustering performance with a silhouette score of 0.4913, successfully grouping data into three main seasonal clusters. Meanwhile, MeanShift produces five clusters but is less effective on small-scale datasets. For regression prediction, Random Forest shows the best performance with an R² score of 0.8921, followed by Linear Regression (0.8402), while SVR produces the lowest performance (0.0077). This research demonstrates that a combination of unsupervised and supervised learning methods can provide a more comprehensive understanding of seasonal patterns and quantitative rainfall estimation. These findings have potential applications in decision-making related to water resource management, agricultural planning, and climate risk mitigation.]]>
Copyrights © 2025
