Enhancement of genetic algorithm applied in e-commerce delivery routing

ABSTRACT: Genetic Algorithm (GA) is an optimization technique inspired by natural selection where a population of candidate solutions evolves over generations through selection, crossover, and mutation. Although GA is effective for solving complex optimization problems, traditional implementations often face challenges like slow convergence and premature stagnation. To address these issues, several enhancements were introduced. The Nearest Neighbor (NN) heuristic was employed during the initialization phase to create high-quality starting solutions by iteratively selecting the closest unvisited point, improving early convergence and reducing the need for excessive exploration. Additionally, Grid Search was used to automate the tuning of key GA parameters such as mutation rate, crossover rate, and population size. By systematically exploring combinations of these parameters, Grid Search identified optimal settings more efficiently than manual tuning, leading to better overall performance. To further support exploration and prevent the population from becoming overly homogeneous. Diversity-Preserving Mechanisms were implemented by dynamically adjusting mutation rates based on measured diversity levels. This approach helped the enhanced GA maintain diversity above 40% across generations, whereas standard GA models typically suffered from diversity loss over time. These integrated enhancements---Nearest Neighbor for smarter initialization. Grid Search for automated parameter optimization, and Diversity-Preserving Mechanisms for sustained variation-----collectively improved the GA’s convergence speed, solution quality, and adaptability. As a result, the enhanced Genetic Algorithm proved more effective and robust in solving complex optimization problems, particularly in real-world applications like E-commerce delivery routing, where maintaining both exploration and solution efficiency is critical.

Filipiniana