Abstract:
In the case of Pakistan the real estate market is very much dependent on manual valuation the agents experience and different levels of market knowledge which are not always consistent. Due to this lack of clarity buyers and sellers get perplexed about the price thereby making it difficult for them to take right decision. To counter this problem an AI driven web portal with a stylish easy to use interface is the project to which accurate property price predictions based on data are given as a solution. The system operates on a twofold backend infrastructure where computational work as well as I/O processes are separately taken care of for superior performance. A key factor of the platform is a Price Prediction Model that utilizes the XGBoost Regressor which is trained on local real estate data to produce accurate automated price estimates and thus helps users to have a clearer understanding of the market value and lowers the reliance on subjective manual evaluations. The application utilizes React.js for the mobile friendly frontend, a dedicated Flask service for quick ML inference and a Node.js/Express backend for administering user authentication API routing and database operations. PostgreSQL is the primary database to ensure that the data is managed in a secure and reliable manner. Tests performed indicate that the XGBoost model has a reliable performance since it has a R2 score of 0.87 on the test dataset. The system gives speedy predictions seamless interaction and a simplified property evaluation process which is why Propex is regarded as a viable digital transformation driver in the real estate sector of Pakistan.
