Statistical Comparison of Multiple Regression Tools in Prediction of the Bond Strength of Bamboo-Reinforced Concrete

In civil engineering, construction materials are valued for being cost-effective, reliable, and environmentally sustainable. Bamboo has become a viable substitute for steel when used in concrete reinforcement. The bonding behavior of bamboo-reinforced concrete up to the point of failure must be understood by predictive modeling in order to lower the high expenses of testing supplies and equipment. In order to forecast the performance of the bond among bamboo reinforcement and concrete, this study demonstrates the potential application of artificial neural networks (ANNs), support vector machines (SVMs), random forests (RFs), Excel, and the Statistical Package for the Social Sciences (SPSS). Excel and SPSS models are used for multiple regression analysis to suggest formulae for bond strength prediction. The predicted results are validated using data from experimental pullout tests. Models were constructed, trained, and tested using a total of 339 pullout test results, of which 215 samples are from bamboo that has been chemically treated and 114 samples are untreated (plain). The study emphasized modeling over the chemical treatment of bamboo. All models utilize bond strength as output and use 12 input parameters. All modeling performs satisfactorily, with a minimum coefficient of determination (R2) of 0.927 for untreated reinforced bamboo (UTRB) concrete samples and 0.970 for treated reinforced bamboo (TRB) concrete samples. The results demonstrate that all models exhibit strong prediction and generalization capabilities with low error rates. However, in this study, the proposed Excel and SPSS models proved to be slightly more accurate than the ANN, SVM, and RF models. Consequently, these predictive models can be employed to quickly estimate the bond strength of bamboo-reinforced concrete without requiring physical tests.