For the purpose of predicting the residual strength of fiber-reinforced concrete when it is subjected to bending stress, the study at hand suggests the use of boosting frameworks. The approaches of \(ML\) known as extreme categorical boosting ( \(CB\) ) are utilized in order to predict the post-peak flexural strength of concrete reinforced with steel fiber ( \(SFRC\) ) at two different levels of crack width. These crack widths include a 0.5 mm crack width ( \({f}_{R,1}\) ) and 2.5 mm crack width ( \({f}_{R,3}\) ). There are two cutting-edge optimization methods that are evaluated for this purpose, and they are named Coati Optimization ( \(CO\) ) and Flood Optimization ( \(FO\) ). The performance of the \(CB\) models is profoundly impacted by their hyperparameters, which may be adjusted via the use of optimization techniques. A total of 216 experimental samples have been gathered for data collection. To prepare the dataset for model introduction, several pre-processing steps are performed. According to the results, it was shown that \(CCB\) and \(FCB\) will provide precise estimates of \({f}_{R,1}\) and \({f}_{R,3}\) . The statistical criteria for performance analysis indicate that \(FCB\) performs better than other models in terms of precision and reliability. These results provide practical guidance for material selection and mix design, particularly for industrial flooring, where residual strength governs crack control and durability. The optimization techniques are employed as supporting tools to enhance model robustness, while the central outcome is an interpretable and reliable assessment of post-peak flexural behavior that complements existing design provisions and experimental testing.
