A multivariate regression to predict daylighting and energy consumption of residential buildings within hybrid settlements in hot-desert climates

Abstract

Over recent decades, Egypt has experienced unprecedented growth of urban residential regions causing deterioration to indoor environmental quality. This research is a part of an ongoing study of building performance with different physical configurations and façades. It aims to quantify the daylighting and energy consumption of residential buildings in the hot-desert hybrid settlements of Alexandria. The methodological approach involves performing computational simulations to construct a dataset covering several influencing factors, which are exploited to develop multivariate linear regression models. This yields five equations used as proxies for predicting building performance in the early stages of the design. The developed models are validated and the predicted data, accompanied by insignificant errors, are found to be in a good agreement with simulation results, indicating that these models can explain the variation in the building performance. These measures, supported by an additional analysis of residuals, confirm the strength of selected variables and suitability of developed models to fit the dataset. This can aid architects and decision makers to assess a preliminary building performance by considering factors of heavily obstructed environment and façade configurations without the need to perform exhaustive analysis.

An occupant-centric approach for spatio-temporal visual comfort assessment and optimization in daylit sports spaces

“…Recently, surrogate models built on machine/deep learning algorithms were extensively employed to handle complex non-linear problems like daylight and glare simulations, 34 serving as proxies to computationally sophisticated simulations. Ayoub et al 35 developed five multivariate linear regression models to predict daylighting and energy consumption of residential buildings in hot-desert climates, which can provide designers with a pre-diagnostic tool without performing exhaustive analysis. Chatzikonstantinou and Sariyildiz 36 utilized machine learning algorithms to predict visual comfort in office buildings; three machine learning methods were compared in terms of their applicability in approximating DA and DGP.…”

Research on Light Comfort of Waiting Hall of High-Speed Railway Station in Cold Region Based on Interpretable Machine Learning

“…Explainable analysis can be used to determine the most important design parameters affecting the light environment of the waiting hall, and designers can quickly quantify the light performance at the early design stage and improve the efficiency of optimizing the light performance of the waiting hall. In order to introduce interpretability into waiting hall light environment prediction, correlation analysis and multiple linear regression equations [24,36] are mostly used to derive the key contributing parameters to the indoor light environment. Although multiple linear regression equations can rank the sensitivity of key parameters of the indoor light environment, they do not consider the nonlinear and complex relationship between explanatory variables and prediction targets, and are prone to multiple covariance.…”

Natural daylight design for underground public space

“…In short, natural light has a better visual effect and thermal effect, which could help to improve the energy efficiency of buildings. 16 To sum up, the full utilization of natural light resources has a practical significance in different aspects for the sustainable development of underground public spaces, as summarized in Figure 1.…”