Decision-making method combining machine learning and expert subjective judgment and its application to typhoon-induced house damage assessment

Abstract

Obtaining high-quality data often poses significant challenges in real-world scenarios, resulting in poorly performing traditional machine learning (ML) models. To address this issue, this study developed a decision-making approach that combines ML with expert subjective examination and applied it to assessing house damage caused by typhoons. First, an ML model was constructed based on similar cases, selecting data from the optimal number of similar cases as the training data, thereby significantly improving data quality. Subsequently, a decision-making method was developed based on evidential reasoning. By integrating the predictive results of multiple ML models, the advantages of various models were utilized to enhance prediction accuracy and robustness. Additionally, expert opinions were integrated to introduce domain knowledge and experience, further optimizing the prediction results. Finally, experiments verified the effectiveness of the proposed decision-making method in evaluating house damage caused by typhoons and compared it with traditional ML algorithms. The results indicate that the proposed method provides a flexible decision-making approach that combines ML and expert subjective examination, thereby effectively enhancing decision accuracy.