APPLICATION OF MACHINE LEARNING IN ANALYSING HISTORICAL AND NON-HISTORICAL CHARACTERISTICS OF HERITAGE PRE-WAR SHOPHOUSES

Abstract

Real estate is complex and its value is influenced by many characteristics. However, the current practice in Malaysia shows that historical characteristics have not been given primary consideration in determining the value of heritage properties. Thus, the accuracy of the values produced is questionable. This paper aims to determine whether the historical characteristics of the pre-war shophouses at North-East Penang Island, Malaysia contribute any significance to their value. Several Machine Learning algorithms have been developed for this purpose namely Random Forest, Decision Tree, Lasso Regression, Ridge Regression and Linear Regression. The result shows that the Random Forest Regressor with historical characteristics is the best fitting model with higher values of R-squared (R2) and lowest value of Root Mean Square Error (RMSE). This indicates that the historical characteristics of the heritage property under study contribute to its significant value. By considering the historical characteristics, the property’s value can be better predicted.