PASSENGER FORECAST DOMESTIC FLIGHTS USING METHOD GATED RECURRENT UNIT

Peramalan Jumlah Penumpang Pesawat Penerbangan Domestik Menggunakan Metode Gated Reccurennt Unit

Abstract

The number of air passengers is a crucial indicator in air transportation planning and management. However, passenger data often exhibits complex characteristics such as seasonality and fluctuations, making accurate forecasting a challenging task. Conventional statistical methods have limitations in capturing nonlinear patterns, thus requiring more advanced approaches. This study aims to develop a forecasting model for domestic air passengers at Kualanamu International Airport using the Gated Recurrent Unit (GRU) method. GRU is selected due to its ability to efficiently model sequential data, overcome the vanishing gradient problem in traditional Recurrent Neural Networks (RNN), and provide a simpler architecture compared to Long Short-Term Memory (LSTM). The dataset consists of monthly passenger data from January 2019 to November 2024. A sliding window approach with a window size of 12 is applied, and the model uses a stacked GRU architecture optimized with Adam and Mean Squared Error (MSE). The results show that the model achieves a Mean Absolute Percentage Error (MAPE) of 10.10% and Root Mean Square Error (RMSE) of 30,121, indicating good forecasting performance. The novelty of this study lies in the implementation of a multi-layer GRU model combined with seasonal feature inputs. The model is further used to predict passenger numbers for the next 24 months, showing consistent seasonal patterns.