The real estate industry is the backbone of the global economy, thus urban development and financial decisions are influenced by this industry. To predict house prices, complex yet highly critical analysis of geography, property features, market trends, and economic conditions is required. Recent advancements in machine learning and regression-based techniques have transformed predictive modelling, enabling models to identify the complex relationships between parameters to improve the performance of the forecasting process. This paper provides a comparative study of regression-based models for the forecasting pricing of houses and application of such models in smart housing, which relies on technology and data to optimize living spaces and strategic decision-making. Among the standalone models, XGBoost delivers a superior performance score of 0.902 R2 and an RMSE of 27,357.97. A hybrid ARIMA-XGBoost model improves the performance with the scores of 0.905 R2 and an RMSE of 27,028.26. The ensemble models also deliver at competitive levels. The findings highlight the addition of temporal and macroeconomic variables for smart housing solutions development and real-estate decision making.