OPTIMIZATION OF TRANSPORTATION NETWORK USING MACHINE LEARNING AND OPERATIONS RESEARCH

Abstract

Rapid development, rising logistical request, and changing traffic situations are all contributing to the difficulty of transportation systems. Conventional Operations Research (OR) methods, mathematical optimization and shortest-path procedures offer ideal answers under static molds but find it tough to adjust to real-time erraticism. On the other hand, machine learning approaches provide forecast skills based on flowing and historical data, but they regularly lack interpretability and conclusion optimality. A detailed literature analysis and mixture of Operations Research (OR), Machine Learning (ML), and hybrid Operations Research-Machine Learning methods for transportation system optimization are accessible in this study. The study highlights datasets, valuation trials, and methodological dissimilarities while critically examining works on routing, congestion organization, cost reduction, and travel-time forecast. According to a relative analysis, Hybrid Operations Research-Machine Learning replicas regularly outclass solo methods, resulting in normal gains in trip time decrease and operating cost reserves of 10–25%. Duplicability, defined assessment measures, and real-world placement validation are still absent, nonetheless. In order to create climbable and repeatable transportation optimization replicas, this research classifies these gaps and delivers methodological proposals. The results suggestion useful information to investigators and experts who want to create dependable, data-driven transportation schemes.