<![CDATA[Accurate passenger demand forecasting is crucial for operational planning and service reliability in public transportation systems. Despite the effectiveness of traditional models, existing approaches often struggle with nonlinear fluctuations in demand, which limits their ability to adapt to real-world variability. This study proposes a hybrid forecasting framework that combines the Autoregressive Integrated Moving Average (ARIMA) model with a Multi-Layer Perceptron (MLP) neural network for short-term passenger demand prediction. By using ARIMA to capture linear components like trend, seasonality, and autocorrelation, and MLP to model the residuals that contain nonlinear patterns, the proposed approach integrates the strengths of both models. This hybrid method addresses gaps in current forecasting techniques by improving adaptability and precision. Empirical analysis was conducted using daily passenger count data from Bus Trans Jatim during 2023–2024. Data preprocessing included exploratory time series analysis, variance stabilization, and outlier assessment to ensure compatibility with the modeling assumptions. Forecast performance was evaluated using the Mean Absolute Percentage Error (MAPE). The results show that the hybrid ARIMA–MLP model achieved a MAPE of 4.95%, outperforming the standalone ARIMA model in providing more adaptive and accurate short-term forecasts. These findings have practical implications for public transportation planning, enabling more responsive and efficient operations, particularly for forecasting demand fluctuations.]]>
Copyrights © 2026
