Flood Hazard Prediction Model Based on K-means Clustering Analysis and AdaBoost Regression Model

In this paper, data analysis and predictive modeling are used as tools to explore the pattern of flood hazard occurrence and provide early warning support. First, indicators with high correlation with flood occurrence, such as infrastructure deterioration and monsoon intensity, are identified through Pearson correlation coefficient, and the results are presented in a visual way. Second, K-means clustering analysis is applied to classify the flood risk hierarchy, and early warning values are provided through linear regression modeling. Finally, a multi-model comparison method is used to select an applicable machine learning regression model, such as the AdaBoost regression model, and an optimal flood probability regression prediction model is designed by combining the key indicators, which provides important support for flood prevention and control.