A comparative assessment of machine learning methods for predicting housing prices using Bayesian optimization

Abstract

The valuation of house prices is drawing noteworthy attention due to worldwide financial and real estate crises in the last decade. Therefore, there is an immediate need to design more effective predictive systems of house prices. Indeed, investors, creditors, and governments are all interested in such predictive systems to improve their buying and lending decisions and activities. This study explores the application of artificial intelligence, machine learning, and nonlinear statistical models to house price prediction problems. In that order, we use boosting ensemble regression trees, support vector regression, and Gaussian process regression. Bayesian optimization is implemented in a ten-fold cross-validation framework to determine their respective optimal kernels and parameter values. Four performance metrics are used to evaluate the prediction ability of each predictive system. The experimental results showed that boosting ensemble regression trees performed the best, followed by Gaussian process regression and support vector regression. In addition, all three aforementioned predictive systems outperformed artificial neural networks and multi-variate regression employed in recent work on the same data set. Under this perspective, it is concluded that boosting ensemble regression trees are clear candidates to be considered for operational house price prediction in Taiwan.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 65 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.