HYBRID ARIMA–ANN MODEL FOR AIR QUALITY INDEX PREDICTION IN DKI JAKARTA

Air pollution is a threat to all countries, including Indonesia. One area in Indonesia with poor air quality is DKI Jakarta. One step to minimize the decline in air quality in an area is to predict the air quality index in the future. In this study, a hybrid ARIMA-ANN analysis was conducted, combining the ARIMA method and Artificial Neural Networks to model air quality in DKI Jakarta. The time series data of the air quality index sourced from the DKI Jakarta Environmental Service during January 19-30, 2023, which was observed every hour with a total of 288 data. The results of the study showed that the SAE and RMSE of the ARIMA model were 94.135 and 1.157, respectively, while the SAE and RMSE values ​​of the hybrid ARIMA-ANN model were 61.094 and 1.15. The results of the study showed that the hybrid ARIMA-ANN model had a higher accuracy value compared to the single ARIMA model in describing DKI Jakarta air quality data. This study has limitations in that determining the network architecture in the ANN model is still done by trial and error, so it takes a relatively longer time.