Peramalan Jumlah Wisatawan Menggunakan Metode Time Series Arima dan Feed-Forward Neural Network (Studi Kasus: Tempat Wisata Pasir Putih, Situbondo)

Abstract

Tourism comes from Sanskrit which means to travel repeatedly. Tourism is the largest industry in Indonesia, since the covid-19 outbreak on March 3 2020, residents around Pasir Putih Beach Situbondo experienced economic problems due to a drastic decrease in the number of tourist visitors. Tourism office and the local management needs to do a fairly large and extensive promotion of the activities that will be held to boost the economy of the surrounding community, the implementation of activities can be based on the highest forecasting value in order to prepare infrastructure to compensate for the surge in the number of tourist visitors. Forecasting is done by implementing ARIMA time-series method and feed forward Neural Network with forecasting accuracy score using Mean Absolute Percentage Error, this study uses monthly tourist visitor data of Pasir Putih Beach Situbondo from January 2017 to December 2021. The results showed that the data used has a seasonal data type with a stationary data pattern, so there is no need to do the process of differencing but need to add seasonal parameters. The best ARIMA model selected is seasonal ARIMA model and the best FFNN model is 1 input layer LSTM model with 72 nodes, 1 hidden layer LSTM with 64 nodes, 1 hidden layer Dense with 18 nodes, and 1 output layer with 1 node with an error value of 4.245956% or has an accuracy of 95.7% The hybrid forecasting value for the number of tourist visitors in 2022 was 16216 visitors in January, 12066 visitors in February, 10357 visitors in March, 7814 visitors in April, 8592 visitors in May, 10110 visitors in June, 13451 visitors in July, 13460 visitors in August, 13466 visitors in September, 13468 visitors in October, 13467 visitors in November, 13465 visitors in December.