Machine Learning, Architectural Styles and Property Values

Abstract This paper couples a traditional hedonic model with architectural style classifications from human experts and machine learning (ML) enabled classifiers to estimate sales price premia over styles, both at the building neighborhood-level. We find statistically economically significant differences for houses distinct styles across an array of specifications modeling assumptions. Comparisons between ML models illustrate conditions under which may perform least as reliable in mass appraisal models. Hedonic estimates that impact on is attenuated by properties less well-defined we no evidence differential effects Revival or Contemporary architecture new construction.