Accurate daily passenger demand prediction is crucial for efficient public transportation operations. Short-term forecasting models offer a valuable tool to address this challenge, particularly in capturing the inherent seasonality patterns that can significantly impact ridership. This study explores the effectiveness of various short-term forecasting models in predicting daily passenger demand, with a focus on real-world data provided by Mowasalat Masr company (March-June 2022). Utilizing exploratory data analysis, model selection, and parameter optimization techniques, we evaluated the performance of three widely used models: Facebook Prophet, SARIMA, and Holt-Winters.
Our analysis revealed that SARIMA emerged as the leading individual model, demonstrating superior ability in capturing seasonality within the data. However, we ventured beyond individual models and explored a hybrid approach combining SARIMA with Holt-Winters. This hybrid model achieved even greater accuracy in forecasting passenger demand, highlighting the potential benefits of combining complementary forecasting techniques.
This study not only identifies a suitable forecasting approach for short-term public transportation demand prediction, but also opens avenues for further research. Future advancements could include incorporating external factors such as holidays or special events that might influence passenger demand. Additionally, exploring more sophisticated hybrid model techniques or applying these methods to larger datasets could lead to even more robust and generalizable forecasting solutions, ultimately contributing to improved public transportation efficiency.