<![CDATA[The significant impact of air pollution, particularly PM2.5, has driven mitigation efforts to reduce health and environmental risks through more accurate prediction systems. In this study, a Deep Learning approach using the LSTM method with the addition of Polynomial transformation features is proposed to predict PM2.5 concentrations. Historical PM2.5 data from South Tangerang City, Banten, was used to train and test the model. The results show that LSTM with polynomial features effectively captures temporal and non-linear patterns in the data, producing accurate and consistent predictions for both training and testing data compared to conventional machine learning methods such as XGBoost and SVR. Polynomial feature transformation significantly improved model performance, as evidenced by the reduction in prediction errors and increased accuracy compared to LSTM without polynomial features. The model also demonstrates adaptability to sudden fluctuations in air quality data. Although the prediction results closely align with actual values, slight discrepancies may arise due to external factors or model limitations. Therefore, the LSTM approach with polynomial feature transformation is an effective and promising method for PM2.5 prediction.]]>
Copyrights © 2025
