Comparative Analysis of Linear and Non-Linear Regression Models in Predicting House Prices in Delta State, Nigeria
Ejiro Stanley Omokoh *
Department of Mathematics, University of Benin, Edo State, Nigeria.
Eduiyovwiri Lewis Ejiro
Department of Mathematics, Delta State College of Education Mosogar, Delta State, Nigeria.
Okogun Augustine Ejiroghene
Department of Mathematics, Delta State College of Education Mosogar, Delta State, Nigeria.
Unaegbu Ebenezer Nkemjika
Department of Mathematics, Nnamdi Azikiwe University Awka, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
This study investigates the comparative performance of linear and non-linear regression models in predicting house prices in Delta State, Nigeria, using a simulated dataset designed to mirror local housing market characteristics. Models evaluated include Ordinary Least Squares Linear Regression, Random Forest, and a feed-forward Neural Network. Performance metrics used are Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R². The analysis includes sensitivity checks on sample size and noise levels, feature importance analysis for tree-based models, and partial dependence interpretation. Findings indicate that non-linear models—particularly ensemble tree-based methods—tend to outperform linear regression in capturing complex interactions among housing attributes, although neural networks require larger datasets and careful regularization to avoid overfitting. Policy implications and recommendations for data collection, model selection, and future research are discussed, with an emphasis on improving housing information infrastructure in Delta State and Sub-Saharan Africa.
Keywords: House prices, linear regression, random forest, neural networks, predictive modeling, delta state, Nigeria, ensemble methods, housing data