Predictive Analytics for Real Estate Valuation Using Machine Learning: A Decision Tree Regression Approach with Grid Search Hyperparameter Optimisation

Predictive Analytics for Real Estate Valuation Using Machine Learning: A Decision Tree Regression Approach with Grid Search Hyperparameter Optimisation presents an advanced machine learning-based system for accurate residential property price prediction across multiple Indian cities. The project integrates automated data ingestion, preprocessing, predictive analytics, and interactive visualization into a complete end-to-end valuation framework. This research evaluates multiple machine learning regression algorithms including Linear Regression, K-Nearest Neighbours (KNN), Support Vector Regression (SVR), Random Forest Regressor, and Decision Tree Regressor using an 8,000-record multi-city real estate dataset. After extensive experimentation and Grid Search Cross-Validation hyperparameter tuning, the Decision Tree Regressor achieved the best performance with 90.0% prediction accuracy, R² score of 0.912, and RMSE of 3.21 INR Lakhs. The proposed system incorporates: Automated data collection using UiPath RPA Multi-stage preprocessing and feature engineering Hyperparameter optimisation using Grid Search CV Interactive Streamlit web application deployment Folium-based geospatial property visualization Scalable and user-friendly real estate valuation architecture The research demonstrates how machine learning can significantly reduce the limitations of traditional manual property appraisal methods by providing objective, scalable, and highly accurate real estate price predictions. The framework is designed for future extensibility with GIS integration, cloud deployment, and deep learning enhancements for temporal market analysis.