Optimizing Building Energy Efficiency with Kolmogorov-Arnold Network (KAN) Backed Regression

With the growing concern for sustainability, this study explores the amagamation of using machine learning to predict energy consumption in buildings. In recent times, due to the continuous increase in urbanization and expansion of the cities, the energy footprint has increased considerably, thereby calling for more efficient architecture to achieve sustainable development. To address this issue, a novel Kolmogorov-Arnold Network (KAN) based regression model is proposed which facilitates the prediction of energy efficiency in terms of cooling and heating loads of buildings. The study uses a diverse dataset that includes features such as relative compactness, surface area, wall area, roof area, orientation, and glazing area for appropriate predictions. The KAN model proves effective in processing the complex relations between the variables with the help of simpler univariate functions. The model has been analyzed for predictive performance using some metrics which include <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex>, MAE, adjusted <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex>, and MSE. The findings suggest that incorporating KAN-based regression into the design process can optimize energy efficiency in buildings, contributing to more sustainable urban planning. This research highlights the importance of advanced machine learning techniques in addressing the complex challenges of energy consumption prediction and supports the development of sustainable and economically viable building designs.