<![CDATA[Based on historical data from BMKG, the Multilayer Perceptron Neural Network (MLP) is used in this study to classify monthly climate patterns in North Sumatra. A machine learning approach is necessary to identify more accurate patterns due to the differences in nonlinear climate elements such as rainfall, temperature, humidity, and air pressure. This research uses a quantitative approach with the MLP model. BMKG data is processed through preprocessing stages, which include monthly feature formation, normalisation, and cleaning. The model is trained with the backpropagation algorithm to optimise the climate pattern classification process. The research results show that MLP is capable of capturing seasonal variations, especially during the transition periods, and can classify monthly climate patterns with high accuracy. Model evaluation indicates stable performance based on accuracy, precision, recall, and F1 score metrics. For classifying monthly climate patterns in North Sumatra, MLP is effectively used. This can also serve as an alternative to the BMKG historical data-based prediction methods that are more suited to nonlinear climate patterns.]]>
Copyrights © 2026
