An A.I. based cost estimation system for material sourcing with product recommendations for architectural designs

ABSTRACT: Accurate early-stage cost estimation is crucial in architectural design, yet it remains a challenge due to various project complexities and client-specific requirements. This study introduces ArchEstimate, an intelligent system that enhances the pre-evaluation and cost estimation process for architectural projects in the Philippines. The system integrates Fuzzy Logic for the initial evaluation phase, taking into account client preferences, project type, and material selection criteria to suggest suitable materials based on predefined rules. In then employs an Artificial Neural Network (ANN) with ReLU activation to estimate the total project cost, using calculated Gross Floor Area (GFA) and selected materials to compute the cost per square meter. In addition, Multi-Criteria Decision Analysis (MCDA) is implemented to support product comparison and provide alternative recommendations based on supplier reliability, quality, and affordability. The system was trained and tested using historical cost data, achieving a precision rate between 90% and 97%, aligning with findings from existing literature that report ANN accuracy between 87% and 98% for similar tasks. The results demonstrate that ArchEstimate significantly improves the accuracy of architectural cost estimation and material selection, serving as a reliable decision-support tool for architects. Future iterations may include collaboration features for subcontractors and expanded databases to enhance usability across broader construction applications. This approach positions ArchEstimates as a valuable innovation in digital construction planning and project managem

Filipiniana