Analysis and Prediction of Road Traffic Incidents in Ibagué Using Artificial Intelligence: An Approach from Industrial Engineering

Abstract

This study addresses road safety in Ibagué, Colombia, using artificial intelligence to analyze and predict traffic accidents. To identify patterns and develop a predictive model based on the Random Forest algorithm, supported by an interactive dashboard for temporal, spatial, and demographic visualization. Historical accident data (2020–2023) was processed through exploratory data analysis (EDA) and class balancing Synthetic Minority Over-Sampling Technique (SMOTE) to reduce bias. The model achieved an F1-score of 0.71, showing moderate effectiveness in classifying accident causes. Young men aged 18–24 were identified as the most vulnerable group, with high motorcycle accident rates during weekend nights, particularly in Comuna 1. Integrating AI and multidimensional analysis enables prioritization of high-risk zones and timeframes. These insights highlight the need for targeted road safety policies, such as educational campaigns and resource optimization in critical areas.