This is a study on asset valuation in the context of Islamic endowments (Awqaf). It uses real data on awqaf properties in the state of Odisha (India) to develop a valuation model. The valuation model can then be used to identify assets that are grossly underpriced and therefore, with maximum return potential from development. The methodology for this study involves data collection, exploratory data analysis, pre-processing of data, and development of models. The study applies various machine learning models such as. K-Nearest Neighbors, Linear Regression, and Random Forest to the dataset and finds an optimal model with 13 variables. with low RMSE and a reasonably high R2 value of 0.617, indicating its robustness. Among the tested models, the Random Forest model with unscaled data and 13 selected variables demonstrated superior performance, and accuracy in predicting real-world asset values. The K-Nearest Neighbors model with scaled data also showed promise with a low RMSE but was less effective overall compared to Random Forest. The study has major implications for the development of awqaf sector in general and Odisha (India) in particular. The undeveloped awqaf assets that are currently not productive but with maximum potential may be short-listed using this model and a comprehensive development plan may be subsequently implemented so as to yield sustainable benefits for intended beneficiaries in the society.