Mar. 2017 – Jan. 2018
With 5 lab members
Funded by Korea Appraisal Board (KAB)
- Analyzed real estate data related to land and house prices in South Korea with Python.
- Applied machine learning algorithms to develop price predicting model using new features and clustering analysis.
Real-estate price prediction is a complex procedure with economical, social, and geographical factors all in play. The traditional method of hedonic real-estate price prediction fails to utilize the benefit of on- going real-estate transaction data and is limited by the subjective nature of the process. The availability of real-estate transaction data accumulated over time provides a circumstance in which machine learning can thrive in. However, a single model machine learning approach can be limited in addressing the high-dimensionality of real-estate data. As a part of a large on-going program, this study evaluates an ensemble-based machine learning approach developed to predict the actual transaction prices of the real-estates located in the city of Daejeon, South Korea using the transaction data accumulated from 2011 to 2016. The results show that the proposed model outperforms a single model machine learning approach, providing valuable insights for future research in real-estate price prediction.