Abstract
Sustainable development might be promoted if waste eggshells are used in cement-based mate-rials (CBMs) by decreasing waste disposal problems, CO2 emissions, and material costs. In this regard, experimental and machine learning (ML) strategies were used to assess the flexural strength (FS) of CBMs comprising waste eggshell powder (ESP). Initially, experimental methods were carried out to evaluate the FS of ESP-based CBMs. The acquired data was utilized to predict the FS by applying ML algorithms like a decision tree and an AdaBoost regressor. The efficacy of ML techniques was measured by comparing experimental and predicted FS and applying statis-tical checks and k-fold evaluations. The experiment outcomes demonstrated that the FS of CBMs was improved by the inclusion of ESP at lower replacement ratios. Additionally, based on the assessment of modeling results, the AdaBoost regressor model demonstrated a higher exactness in estimating the FS of CBMs containing ESP compared to the decision tree model. The assessment of error readings also indicated the higher precision of the AdaBoost regressor model than the de-cision tree. According to the feature importance analysis, the fine aggregate was the component with the lowest impact for FS, whereas cement had the prime impact. The findings showed that as opposed to using ESP as a cement replacement, using ESP as an FA replacement to keep FA in less quantity and cement in larger quantity would result in enhanced strength. The construction in-dustry can gain from ML techniques since they allow for more time-and cost-effective approaches to assessing material attributes.