Application of Autoregressive Neural Network (ARNN) Hybrid to Forecast the Covid-19 Vaccinations in Java and Bali

Abstract

The Covid-19 virus that has swept the world requires researchers to develop vaccines to prevent transmission. In Indonesia, vaccine distribution seems slow due to the spread of hoaxes on social media and the lack of supply of vaccine doses. Forecasting the number of vaccination distributions is expected to help accelerate the formation of herd-immunity. Forecasting is carried out using the ARNN Hybrid method on daily Covid-19 vaccination data for 7 provinces on the islands of Java and Bali in the period January - December 2021. A quantitative approach is a suitable approach for predicting with a systematic sampling technique. Forecasting with ARNN produces the best model with estimated results that are similar to actual data and low MAPE values. Forecasting with ARNN is considered very good in predicting Covid-19 vaccination activity on the islands of Java and Bali. However, the research in this thesis has not described a clear detrending and deseasonalized procedure because of the limited literature available. Therefore, the author suggests conducting similar research with other methods and the need for further development of ARNN.