<![CDATA[This study is an experimental and comparative quantitative research that evaluates missing value imputation methods for daily rainfall data in West Java. Rainfall data are crucial for environmental policies, particularly in flood control and water resource management. Daily rainfall records from five BMKG stations in West Java were used in this study. Although these stations provide accurate data through direct measurement, missing values often occur due to human error or equipment problems. To solve this, we introduce an integrated imputation method that combines K-Nearest Neighbors (KNN) and Singular Value Decomposition (SVD) with a Weighted Linear Combination (WLC) approach. This method represents a significant improvement over the single-model imputation methods employed in earlier research. We split the dataset into training and testing sets using five different ratios (95:5%, 90:10%, 80:20%, 70:30%, and 64:40%) to test the model's performance. We measured effectiveness using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The results show that the combined KNN–SVD method outperforms KNN or SVD alone in all cases. The best results were obtained from the 95:5% split, with the lowest MAE and RMSE values of 7.35 and 13.22, respectively. These results suggest that the integrated KNN–SVD imputation model enhances the reliability of rainfall datasets, thereby improving climate information for hydrological studies, disaster risk reduction, and policy-making in West Java.]]>
Copyrights © 2026
