Fast and Accurate Estimation of Building Cost Using Machine Learning

Accurately predicting building cost is of a great importance to house building companies. In this study, a machine learning (ML) framework with several regression approaches is developed to model and estimate the building costs accurately in both actual and inverse ways. A dataset of 10,000 samples, the real data computed by different services of our partner construction company, is used to train and to validate ML models. The ML-based estimated results indicate that linear regression model and decision tree model provide the most accurate results for construction cost and maintenance cost, respectively. Furthermore, an artificial neural network framework is considered in the inverse way to get the highest regression accuracy in order to identify the best available features of to-be-built house that a buyer can have for a given budget.