Multiobjective project clustering optimization for enhancing highway contract bundling decisions

Project bundling, which combines multiple projects into a single contract, has emerged as a strategic approach for state highway agencies (SHAs) to improve efficiency and reduce costs in highway construction. However, due to the lack of systematic methods for evaluating bundling candidates, most SHAs face challenges in determining which projects should be bundled together. While several quantitative methods have been proposed for project bundling based on project scope similarity, they failed to consider the spatial proximity of projects, which significantly impacts the effectiveness and feasibility of project bundles. This study addresses this gap by introducing a novel multiobjective clustering method that incorporates both project scope similarity and spatial proximity. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) algorithm was used to optimize the clustering of projects, simultaneously maximizing both scope similarity measured by the cost-weighted term frequency-inverse document frequency (CW-TF-IDF) and spatial proximity. The application of the proposed method was demonstrated on historical contract bundles of the Iowa Department of Transportation. The results of this study indicate the effectiveness of the proposed multiobjective clustering approach that shows a high agreement level with human experts, achieving a v-measure of above 0.6. This study offers SHAs with a new objective project bundling technique that is expected to enhance bundling decisions and the efficiency and cost-effectiveness of highway construction projects.