PERAMALAN TRANSAKSI PEMBAYARAN NON-TUNAI MENGGUNAKAN ARIMAX-ANN DENGAN KONFIGURASI KALENDER

Putera, Muhammad Luthfi Setiarno

BAREKENG: Jurnal Ilmu Matematika dan Terapan
Vol 14 No 1 (2020): BAREKENG: Jurnal Ilmu Matematika dan Terapan

Huge internet usage boosts transactions using non-cash payment. In Indonesia, lots of activities and transactions are influenced by calendar movement, particularly that of the Islamic calendar. This work is to obtain the model and to forecast total non-cash payment with calendar configuration as an exogenous variable. The methods being compared are ARIMA, ARIMAX, and hybrid ARIMAX-ANN. The best model to forecast a total of non-cash payment is ARIMAX-ANN due to the least RMSE, Rp 20,9 trillion. The specification of the best model is ARIMAX (2,1,1) combined with ANN whose input is selected through stepwise regression. Besides satisfying residual assumption, ARIMAX-ANN is quite well in capturing the dynamics and trend of non-cash payment, particularly that in Ied-Fitr month and end of the year.