PREDICT WHETHER THE BUILDING FOLLOW’S GREEN NORMS OR NOT USING MACHINE LEARNING: AN OVERVIEW

Abstract

In this study, we investigate the determinants and predictive models for the green ratings of buildings, aiming to discern patterns and factors that influence whether a building is classified as "legal" or "illegal" in terms of environmental sustainability. Utilizing a comprehensive dataset on green buildings, we first clean and preprocess the data to handle missing and duplicate values, followed by an extensive exploratory data analysis (EDA) to understand the underlying distributions and correlations among features. We employ various machine learning techniques, including Logistic Regression, Decision Tree Classification, and Linear Regression converted to a classification task, to build predictive models. The models were evaluated on the basis of accuracy, precision, recall, and F1 scores. Our findings are visualized using correlation and confusion matrices, highlighting the performance and reliability of each model. In addition, we calculate the Root Mean Squared Error (RMSE) to assess the regression model’s performance when adapted for classification purposes. Overall, this study contributes to the field of sustainable building practices by providing a detailed analysis and comparison of machine learning models for green rating predictions, offering insights and tools that can aid policymakers, builders, and environmental analysts in promoting and adhering to green building standards.