A data-driven approach to building design and contract pricing using explainable machine learning

This study aims to quantify architectural design features and predict and explain their relationship with contract prices. This work focuses on the random forest algorithm, one of the machine learning algorithms. This algorithm is enhanced through parameter optimizations and compared with Shapley Additive Explanations (SHAP) to measure each architectural feature's contribution to the model's prediction. The combined evaluation of feature importance and SHAP significantly enhanced the model's interpretability. The results showed that the roof area and the number of doors had a significant impact on project costs among the building's architectural features. The performance evaluation metric outputs obtained, with R2 at 0.8725, NSE at 0.8709, MAE at 0.6510, and RMSE at 0.9615, align with the developed prediction model. The study reveals the relationship between contract costs and their sub-variables using predictive models and explainable machine learning methods to build architectural features. Copyright (c) 2026. Aslay, S. This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivatives 4.0 International License https://creativecommons.org/licenses/by/4.0/.