Machine learning models based web application for predicting flexural strength of steel fiber-reinforced concrete

This article presents the performance of two predictive learning techniques, including k-nearest neighbor (k-NN) and multiple linear regression (MLR), for calculating the flexural strength of steel fiber-reinforced concrete (SFRC). To achieve this, a set of 92 experimental results on the flexural strength of the SFRC were carefully collected from the literature reviews and used to train and test the predictive models. The coefficient of determination ( R 2 ) and root mean square error ( RMSE ) a were employed to assess the performance of the k-NN and MLR models. The analytical outcome proved that both the k-NN and MLR models exhibited good predictions for the flexural strength of the SFRC. The k-NN model displayed higher accuracy than the MLR model, with R 2 and RMSE values of 0.83 and 1.09 MPa for training and 0.77 and 1.28 MPa for testing, respectively. Moreover, a simple website was designed to help quickly predict the flexural strength of the SFRC for practical employment.