Web Application for Passenger Demand Prediction in a Public Transportation Company in Metropolitan Lima Based on Random Forest

Villanueva W.; Torero M.; Wong L. · Applications in Software Engineering - Proceedings of the 14th International Conference on Software Process Improvement, CIMPS 2025 (2025) · DOI: 10.1109/cimps69356.2025.11419687

Resumen

Public transportation in Metropolitan Lima faces serious operational deficiencies due to the lack of technological tools capable of accurately predicting passenger demand, resulting in economic losses and resource overuse. This study developed a web application based on the Random Forest algorithm to predict passenger demand for a public transportation company in Peru. A real dataset with over 3.9 million records was used, including variables such as bus stop, time, and number of passengers. Validation was carried out through two real-world experiments under operational conditions. The first used a traditional approach without technological support; the second utilized the predictive application. In the second experiment, the average bus occupancy increased by 36.72 %, unnecessary buses were reduced by 77.78 %, and economic losses also decreased by 77.78 %. These metrics demonstrated a substantial improvement in operational efficiency. The results showed that the use of Random Forest-based prediction significantly improves resource allocation, reduces costs, and optimizes strategic decision-making in urban public transportation companies, establishing the proposed solution as an effective, scalable tool aligned with the efficiency objectives of the sector.

Instituciones: Universidad Peruana de Ciencias Aplicadas