This study explores the use of rice husk ash (RHA) as a supplementary material in cement mortar, focusing on its impact on compressive strength through machine learning predictive modelling. The methodology involved a systematic literature review to compile a comprehensive dataset of 692 records from 20 published sources. A range of machine learning techniques, comprising Linear Regression, Artificial Neural Networks, Random Forest Regression, and Extreme Gradient Boosting, were utilized to assessment compressive strength based on important variables such as the ratios of aggregate-to-binder, RHA-to-binder, water-to-binder, along with the curing time and chemical composition. Results designate that RHA significantly contributes to the pozzolanic activity of cement mortar, with optimal RHA-to-binder ratios enhancing compressive strength without compromising workability. The machine learning models exhibited high predictive precision, with R² values more than 0.95 and low RMSE values across the tested datasets. Sensitivity analysis has shown that the aggregate-to-binder ratio, SiO2 content and curing period were the important parameters affecting compressive strength, emphasizing the significance of precise formulation. The study recommends further research to validate the predictive models through experimental studies, broaden the dataset for enhanced generalizability, and explore additional chemical compositions of RHA.

cement mortar, machine learning, oxide, compressive strength