Perbandingan ARIMA dan Artificial Neural Networks dalam Peramalan Jumlah Positif Covid-19 di DKI Jakarta

Abstract

DKI Jakarta menjadi pusat penyebaran Covid-19. Hal ini ditunjukkan dari jumlah kumulatif positif Covid-19 di DKI Jakarta lebih tinggi dibandingkan dengan provinsi lain. Tingginya kasus di DKI Jakarta menjadi perhatian bagi semua kalangan, sehingga perlu dilakukan peramalan untuk meramalkan jumlah positif Covid-19 periode selanjutnya. Peramalan yang akurat dibutuhkkan untuk mendapatkan hasil yang lebih baik. Penelitian ini membandingkan metode Autoregressive Integrated Moving Average (ARIMA) dan Artificial Neural Networks (ANN) dalam peramalan jumlah positif Covid-19 di DKI Jakarta. Akurasi peramalan dihitung menggunakan nilai Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), dan korelasi. Hasil penelitian menunjukkan bahwa model terbaik untuk peramalan jumlah positif Covid-19 di DKI Jakarta adalah ARIMA(0,1,1) with drift dengan nilai MAPE sebesar 15,748, RMSE sebesar 268,808, dan korelasi antara nilai ramalan dengan nilai aktual sebesar 0,845. Peramalan menggunakan model ARIMA(0,1,1) with drift dan BP(3,10,1) menghasilkan ramalan terbaik pada panjang periode peramalan enam minggu ke depan. Kata kunci: ANN, ARIMA, Covid-19, peramalan

DKI Jakarta is the center of the spread of Covid-19. This is indicated by the higher cumulative number of Covid-19 positive in DKI Jakarta compared to other provinces. The high number of cases in DKI Jakarta is a concern for all groups, so it is necessary to do forecasting to predict the number of Covid-19 positive in the next period. Accurate forecasting is needed to get better results. This study compares the Autoregressive Integrated Moving Average (ARIMA) and Artificial Neural Networks (ANN) methods in predicting the number of Covid-19 positive in DKI Jakarta. Forecasting accuracy is calculated using the value of Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and correlation. The results show that the best model for forecasting the number of Covid-19 positive in DKI Jakarta is ARIMA(0,1,1) with drift, with a MAPE value of 15.748, an RMSE of 268.808, and the correlation between the forecast value and the actual value of 0.845. Forecasting using ARIMA(0,1,1) with drift and BP(3,10,1) models produces the best forecast for the long forecasting period of the next six weeks. Keywords: ANN, ARIMA, Covid-19, forecast