Enhancing Compressive Strength Prediction of Calcined Clay Cements with Data Augmentation in Linear Regression Models

Abstract

Cement production is a major contributor to global CO2 (Carbon dioxide) emissions. To minimize its 
environmental impact while maintaining the required mechanical properties of cement, there is a pressing need for 
sustainable pro- duction processes. This paper focuses on developing sustainable cement produc- tion processes by 
optimizing the mechanical properties of limestone calcined clay cement (LC3) using data-driven models based on 
artificial intelligence. The study explores the use of data augmentation techniques, specifically the copulas method, 
to improve the performance of linear regression models for linking the compressive strength of LC3 with its mix 
design. While data augmentation using copulas can be useful in augmenting tabular data, its effectiveness in 
improving linear regression performance may depend on the statistical characteristics of the original data. The 
method successfully generated additional data that preserved the original statistical properties, but it did not always 
lead to significant improve- ments in linear regression performance. The research highlights the potential of data
driven models for optimizing cement materials properties and emphasizes the importance of considering the 
statistical characteristics of the original data when applying data augmentation techniques