Estimating the Cost of Building Reinforced Concrete Homes Using Artificial Intelligence Models

Estimating construction costs constitutes a strategic component of project management, because it provides fundamental information for assessing its economic feasibility. The present research aims to develop artificial intelligence models for estimating the construction cost of reinforced concrete housing. The database was constructed from 276 technical files corresponding to the period 2019–2025. Random Forest achieved the highest average R2 across the three input-variable groups (R2 = 0.8520), followed by XGBoost and Multiple Linear Regression, both with an average R2 of 0.8473, whereas MLP Regressor obtained a lower value (R2 = 0.8197). Likewise, it presented the lowest average standard deviation across the three groups (σ = 0.0530), compared with XGBoost (σ = 0.0547), Multiple Linear Regression (σ = 0.0719), and MLP Regressor (σ = 0.1023). Compared with Groups 1 and 2, the Random Forest model, using Group 3, presented lower MAE values by S/ 3174.91 and S/ 3185.34, respectively, while the MAPE values were lower by 4.19 and 4.40 percentage points, respectively. The main scientific contribution of this research lies in establishing a reproducible methodology, which made it possible not only to identify the model with the highest predictive capacity but also to characterize the conditions under which its accuracy decreases.