House price prediction in Ajah, Lagos using a random forest regression model

Abstract

Housing is one of the most essential need to commoner but several factors contribute to acquiring a suitable home,  factor such as low income, high material cost and most importantly the presence of middleman in acquiring suitable house posed significant challenges thereby increases the cost, this study addresses this problem by developing a random forest regressor model to predict house price in Ajah Lagos, a dataset of 24,326 instances was sourced from Kaggle and filtered to 2142 instance containing Ajah listing. The data underwent several preprocessing steps; outliers were detected using the interquartile range, and Recursive Feature Elimination was used to identify relevant features that contribute to the model while eliminating less important ones. The RF model demonstrates a robust performance with an R-squared of 68.5%, a Mean Absolute Percentage Error of 18.96%, and a Root Mean Squared Error of ₦11.1 million. Compared with existing models in the literature, the model achieved a mean absolute error of ₦8.09 million, representing a 4.82% improvement and outperforming the ridge regression models. This research contributes to the field by providing stakeholders with a data-driven tool to enhance transparency and investment decision-making in the Nigerian real estate market.