Bayesian Networks Prediction of Compressive Strength of Recycled Aggregate Concrete
Abstract
Many studies proposed machine learning approaches for prediction models analysing the impact factors on recycled aggregate concrete (RAC) compressive strength. However, most machine learning algorithms require a large dataset size for the model's generalisation capability. Few studies have used Bayesian Networks (BNs) based probabilistic inference techniques towards this aim. This paper uses BNs to predict the compressive strength of recycled aggregate concrete. The BNs approach utilised available data of three input parameters: water-to-cement ratio, aggregate-to-cement ratio, and recycled aggregate replacement ratio to compute the output's prior and posterior probability of RAC's compressive strength. The results highlight the potential applicability of BNs in predicting the compressive strength of RAC.