Exploration of Machine Learning Algorithms for Housing Price Prediction in Megacities

Abstract

Effective predictive models are good for tackling pressing issues faced by residents and stakeholders regarding the cost of houses in cities. Thus, buyers and sellers of houses have challenges in making decisions due to a lack of adequate data-driven decision-making. This study explores the feasibility of using machine learning techniques to predict house prices in Lagos and Abuja, Nigeria. The data used for this research were obtained from a property listing website and encompass key features such as location, number of rooms, property type and other basic characteristics. Several models including Linear Regression, Bayesian Linear Regression, Support Vector Regressor and Ridge Regression, were trained using these data. The performances of models considered for this study were evaluated using the Mean Absolute Error and the R-Squared scores, and the Support Vector Regressor performed best. Also, the model predicted Lagos house prices better than Abuja house prices with the highest R-squared and lowest Mean Absolute Error (MAE) values. An independent t-test was also conducted to test for significant differences in the price of houses in Lagos and Abuja. The results indicate there is no significant difference in the prices. The study concludes that machine learning is a useful tool for house price prediction to its accuracy in predictions, highly dependent on the availability and quality of comprehensive datasets.