Performance of Levenberg-Marquardt Neural Network Algorithm in Air Quality Forecasting

Levenberg-Marquardt algorithm and conjugate gradient method are frequently used for optimization in multi-layer perceptron (MLP). However, both algorithms have mixed conclusions optimizing MLP time series forecasting. This study uses autoregressive integrated moving average (ARIMA) with method. These methods were to predict the Air Pollutant Index (API) Malaysia's central region where represent urban residential areas. The performances discussed compared using mean square error (MSE) absolute percentage (MAPE). result shows that models outperformed ARIMA