The study proceeded from the constitutional and tax-law framework governing mass real estate valuation—legality, equality, ability-to-pay, proportionality, and the non-confiscatory principle—together with the requirements of reason-giving and transparency that underlie CAMA/GIS systems and quality assurance. On that basis, it set out to determine the extent to which the 2025 official assessed value in the Historic Center of Lima could be reconstructed from observable attributes; correlatively, it advanced the hypothesis that such assessed value would be broadly explainable without resorting to its accounting components, provided those principles were effectively observed. Using a cleaned universe of approximately 441,724 records, the study implemented a reproducible pipeline (imputation, standardization, and one-hot encoding) and compared three approaches: ordinary least squares (OLS), Ridge, and Random Forest. The latter achieved the strongest explanatory performance, with \(R^2 \approx 0.9667\) , \(\textrm{RMSE} \approx 15{,}150\) , and \(\textrm{MAE} \approx 5{,}883\) , alongside centered residuals and a y– \(\hat{y}\) cloud closely aligned to the \(45\circ \) diagonal; however, error increased in the upper area quintiles, suggesting residual heterogeneity linked to scale effects and incomplete descriptors of the built environment. The principal contribution was to show—using open data and an auditable procedure—that the tax base was largely verifiable by physical and administrative traits, thereby enabling objective criteria to prioritize audits and to strengthen the Statement of Reasons in administrative decisions. Future research should incorporate geospatial and zoning layers, enrich construction-quality descriptors, explore gradient boosting and interpretability (e.g., SHAP), conduct systematic hyperparameter searches and spatial cross-validation, and monitor drift and equity across subpopulations.