Paper Title
Application of Machine Learning Models for the Compressive Strength Prediction of Concrete with Glass Waste Powder

Abstract
In Portland cement concrete, recycled waste glass is frequently utilized as aggregate or supplemental cementitious material. One of the most important properties of concrete, compressive strength, depends on pozzolanic reactivity of glass powder which is influenced by the size of glass particles. In order to incorporate glass waste powder in concrete, a prediction model that will provide the compressive strength of such concrete with a reasonable level of accuracy is needed. In order to develop the best prediction model, this study takes into account a variety of machine learning models built on a dataset of 70 experimentally tested samples of concrete with waste glass. Considered are models based on Gaussian process regression (GPR), Regression Tree (RT), and ensemble models: Random Forest (RF) and Tree Bagger (TB). Analyses are done on the precision of both individual and ensemble models. The most accurate model is presented by the research. Keywords - Concrete with glass cullet, Compressive strength, Gaussian process regression, Random Forest, Tree Bagger.