Real Estate Prices
This project aimed to enhance property valuation accuracy for condominiums in Dallas using a data-driven approach. We developed a predictive model using multiple linear regression, leveraging key features such as Bedrooms, Bathrooms, Square Footage, Lot Size, and Property Type. Despite not achieving the target R² value of 0.9, the analysis revealed that Square Footage is the strongest predictor of property prices. The project highlights the potential of machine learning in improving real estate valuations and provides insights for future model enhancements.
Business Case
In the bustling real estate market of Dallas, this assignment undertakes a data-driven approach to enhance property valuation accuracy. Focusing exclusively on condominiums, our goal is to construct a predictive model utilizing advanced analytics and machine learning. By leveraging features such as Bedrooms, Bathrooms, Square Footage, Lot Size, and Property type, our aim is to create a robust model that exceeds industry standards, with a minimum R2 value of 0.9.
The primary objective is to build a predictive model using multiple linear regression, incorporating key features such as Bedrooms, Bathrooms, Square Footage. The model is expected to exhibit high predictive accuracy, with a minimum R2 value of 0.9, ensuring its reliability and relevance for real-world applications.
Linear Regression Metrics:
| Validation | Cross-Validation |
R-Squared | 0.7645 | 0.6367 |
MAPE | 58.27% | 71.78% |
MAE | $219,160 | $312,730 |
RMSE | $346,470 | $568,320 |
R2 suggests an unacceptable level of performance of the model since the acceptable value for R2 is 0.9. We can also see MAPE, RMSE and MAE. Thus, this model is unlikely to offer value to practitioners.