A comprehensive analysis of optimization algorithm performance in regression and classification scenarios using transportation data

Accurate prediction and classification of transportation-related metrics such as CO 2 emissions and road fatalities are vital for sustainable urban planning and public safety. Using data from 30 major Chinese cities, this study introduces a comprehensive methodology that integrates the XGBoost algorithm with 16 distinct metaheuristic optimization techniques to compare algorithms performance in both regression and classification tasks. Each optimizer is benchmarked based on accuracy, best convergence, and runtime, utilizing a composite score that integrates these parameters for a comprehensive performance evaluation. The findings of this study rank optimization algorithms in terms of accuracy, runtime, and convergence. Results show that AOA and GWO achieves the highest rank for regression tasks while in classification tasks, SMA and AEO demonstrate the best accuracy. This assists researchers in selecting the most suitable optimization algorithm for their specific tasks based on balanced performance criteria.