Machine Learning-Based Search Strategy for Water Object Retrieval in Cultural Tourism Safety Contexts

Abstract

This research addresses the challenge of predicting deviations in the landing positions of objects dropped into water, with important implica-tions for cultural tourism safety near lakes, rivers, and other natural attractions. An innovative optimization method for search strategies based on machine learning is proposed. A simulated dataset incorporating features such as drop height, water entry angle, drag coefficient, and object density enables detailed model comparisons. Five machine learning models—XGBoost, Random Forest, Decision Tree, Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP)— are evaluated using Mean Squared Error (MSE), Mean Absolute Error (MAE), and the Coefficient of Determination. Experimental results show that XGBoost significantly outperforms the others, effectively capturing complex nonlinear relationships through its gradient boost-ing mechanism. In contrast, models like Decision Tree, SVM, and MLP exhibit lower predictive accuracy due to weaker generalization capabilities. This study provides a robust machine learning-based framework to enhance predictive accuracy and search efficiency in aquatic environments.