Comparación de modelos de predicción para estimar la demanda de viajes en radio taxis

The radio taxi services sector constantly faces the challenge of managing variable travel demand, underscoring the importance of using predictive models to optimize resources and enhance service quality. This article focuses on conducting a comparative analysis among three models: ARIMA, Prophet, and Random Forest. Robust evaluation measures are employed to identify the most suitable approach for accurately predicting travel demand in the Radio Taxi service of Cooperativa Rápido Nacional. The results reveal that ARIMA significantly outperforms the other models, exhibiting a Mean Absolute Error (MAE) of 1.46, Mean Squared Error (MSE) of 4.71, and Root Mean Squared Error (RMSE) of 2.17. This demonstrates superior consistency and precision in its predictions compared to the Prophet (MAE: 2.83, MSE: 14.62, RMSE: 3.82) and Random Forest (MAE: 3.27, MSE: 17.03, RMSE: 4.12) models. This analysis highlights the effectiveness of ARIMA in predicting travel demand in radio taxi services, providing valuable insights for enhancing planning and operational efficiency.