Predicting house prices is known to be one of the most challenging tasks in real estate, for it has a heavy impact on the decisions of buyers, sellers, and investors. In this work, we propose a holistic approach that automates the collection and analysis of real estate data to generate accurate price predictions. It utilizes web scraping with Selenium to obtain property listings on various websites and supplements them with spatial data from other APIs. This data collection framework goes through feature extraction, data collection, and training the system with strict pre=processing steps, which include feature engineering, normalization, and missing value treatment to ensure it is ready for model training. To determine the most accurate prediction technique, a number of regression methods are tested, such as: Linear Regression, Gradient Boosting, LightGBM, Ensemble of RF and GB (voting), and Random Forest. The experimental results of this work confirms that the accuracy of the Random Forest model is noticeably better than all other models, thus making it the optimum choice when deployed. Further, this work analyzes how property descriptions can impact pricing and seeks to find ways of deriving value from unstructured data. In hope, this paper gives deeper understanding of e-Commerce architecture, its functioning and practical applicability in engineering of information systems in e-Commerce system for making property value estimation simpler and more efficient.