The National Railway Company of Belgium (NMBS/SNCB) faces the challenge of efficiently managing its dense rail network while ensuring reliable, safe, and comfortable services for its passengers. To address this challenge, this project proposes a data-driven approach to predict train occupancy levels and optimize maintenance scheduling. By leveraging historical data on train schedules, occupancy, weather conditions, and demographic information, the project aims to develop a predictive model that can accurately forecast future occupancy levels across different stations and lines. The project employs a combination of spatial and temporal analysis techniques to uncover patterns and relationships among various factors influencing train occupancy. Through feature engineering, the model incorporates variables such as the proximity of nearby stations, station density, temporal lags, and rush hour indicators to capture the complex dynamics of passenger demand. The modeling approach utilizes Multinomial Logistic Regression to predict occupancy levels categorized as low, medium, or high. The model's performance is evaluated using metrics such as accuracy, sensitivity, specificity, and cross-validation techniques to ensure robustness and generalizability. The ultimate goal is to provide actionable insights for the NMBS/SNCB to optimize maintenance scheduling, minimize service disruptions, and improve overall operational efficiency and passenger satisfaction.
.png)