<![CDATA[Land expansion for residential and industrial areas in a city causes an increase in temperature and humidity. Temperature and humidity changes can be predicted using dynamic spatial modeling as part of the urban planning process. This study aims to predict surface temperature dynamics in Semarang using an Artificial Neural Network (ANN) algorithm, including training, testing, and prediction processing. The data sources used in this study were changes in temperature and humidity prediction factors for 2025 from the BMKG Climatology Station, Central Java UPT. The network architecture model used 2-15-1, and the MSE (Mean Squared Error) reached 0.113. The ANN modeling prediction output showed an average surface temperature prediction of 27.77°C. When compared with the actual value, this model had a fairly good R value (correlation coefficient) during training, with 89% of the data variation. Based on the data, the model is categorized as good and suitable for use in prediction cases.]]>
Copyrights © 2025
