Development of A Machine Learning-Based Model for Predicting Compressive Strength of Ordinary Portland Cement

Cement is a vital construction material with widespread use in the construction industry, acting as a binding agent for various construction materials. The compressive strength of cement, which measures its binding force and ability to withstand compression, is a crucial factor in manufacturing cement and constructing concrete-based structures. Traditionally, costly laboratory tests have been employed to determine cement's compressive strength. However, with the complexity of material engineering, this approach has become inefficient, leading to resource and time losses. Establishing a logical connection between cement's chemical composition, physical characteristics, and compressive strength is also challenging due to its heterogeneous properties and nonlinear behaviour. However, to address these issues, with the evolution of machine learning and its efficient modelling techniques, different modelling techniques are prepared to study its behaviour and satisfy the desired performance. This paper aims to demonstrate the effectiveness of different shallow supervised machine learning techniques such as Multivariant linear regressions, Decision Tree (DT), Nonlinear regression, and ensemble Random Forests (RF) and apply Principal Component Analysis (PCA) to develop a compressive strength prediction model to overcome the disadvantages of traditional approaches (experimental analysis) in estimating the compressive strength of cement. Finally, the study has compared the results obtained by these different Machine Learning (ML) techniques and provided a general conclusion.