Machine learning prediction of compressive strength for phase change materials integrated cementitious composites

Incorporating phase change materials (PCMs) into cementitious composites has recently attracted paramount interest. While it can enhance thermal characteristics and energy storage, compressive strength can be decreased. Thus, accurate prediction of the effect of PCM addition on compressive strength is crucial. However, a predictive model for this purpose using physical or chemical features is not feasible at this stage. Thus, machine learning is used for the first time herein to predict the compressive strength of PCM-integrated cementitious composites. A dataset of 154 cement-based mixtures incorporating PCM microcapsules was assembled. Various machine learning regression algorithms including random forest, extra trees, gradient boosting, and extreme gradient boosting were tuned and their prediction accuracy was assessed using several metrics. The models achieved superior prediction accuracy. Exploiting powerful machine learning models to examine the harvested experimental data could provide insights into materials science aspects of this problem and identify pertinent knowledge gaps and needed future research.