PSI - Issue 68

Hamidreza Rohani Raftar et al. / Procedia Structural Integrity 68 (2025) 1066–1073 Hamidreza Rohani Raftar et al./ Structural Integrity Procedia 00 (2025) 000–000

1069

4

!"#" $%&&D$#(%)*

!"#$%&'(

Fig. 1. Data gathered for training and testing purposes to develop a predictive model using machine learning techniques

2.2. Machine learning method selection and evaluation The selection of machine learning methods to address a problem is a crucial decision across various contexts. In this study, several methods were employed, and the best-performing one was chosen based on error metrics. The collected data underwent a data cleaning procedure to mitigate noise and enhance results, ensuring that filtration was smooth yet preserved the accuracy of the data's underlying patterns. The primary method utilized was bagged trees. This approach was chosen based on its effectiveness in minimizing error metrics and achieving optimal predictive performance. Additionally, several other methods were explored to construct predictive models, including neural networks, trees, linear regression, kernel methods, and support vector machines (SVM), among others. Each method was evaluated for its suitability and contribution to the study's objectives. The error metrics for methods are provided in Table 1, with bagged trees showing notable performance in predictive accuracy.