<![CDATA[Air pollution is an environmental problem with important implications for public health. Particulate Matter 2.5 (PM2.5) is a critical air-quality parameter because its fine particles can penetrate deeply into the respiratory system. This study develops a Long Short-Term Memory (LSTM) model to predict daily PM2.5 concentrations in Semarang City using historical observations. The research stages include preprocessing, Min-Max normalization, sequence generation with a 30-step time window, chronological 80:20 data splitting, LSTM training, and evaluation using MAE, RMSE, and MAPE. The model consists of two LSTM layers with 64 and 32 units, dropout of 0.2, the Adam optimizer, and MSE loss. The testing results produced an MAE of 5.7560, an RMSE of 7.4613, and a MAPE of 16.73%. These results indicate that LSTM can capture temporal patterns in PM2.5 data, although further evaluation using benchmark models and meteorological variables is needed.]]>
Copyrights © 2026
