Machine Learning-Based Prediction of Acoustic Indicators for Enhanced Educational Building Comfort

The dataset comprises two types: numerical data and heat maps. The numerical dataset includes extensive parametric simulations of acoustic indicators in different geometric configurations using the Grasshopper and Pachyderm plugins. These datasets were utilized to predict both numerical and visual acoustic indicators using the CatBoost algorithm and the pix2pix algorithm. The results demonstrate remarkable accuracy, ranging from 89% to 99% in the numerical portion using CatBoost, with PSNR indices ranging from 0.817 to 0.970, and SSIM indices ranging from 15.56 to 31.57 in the visual section.