Optuna-optimized GBDT approach for soil stratification analysis in cone penetration testing

Traditional soil layer classification methods primarily rely on the combination of in-situ testing and expert evaluation, which is often associated with high costs and subjectivity. In order to overcome these deficiencies in traditional soil stratification identification, we propose an algorithm of gradient boosting decision tree(GBDT) based on Optuna hyperparameter optimization framework for soil classification prediction.. Cone penetration test(CPT) data measured in five Yellow River fluvial plain cities in Shandong Province were collected, then divided into training sets and testing sets, finally the prediction results were evaluated by four indicators of accuracy rate, precision rate, recall rate and F1 score. The evaluation results were compared with results of other algorithms such as KNN, RF, MLP, XGBoost, LightGBM, CatBoost, and the proposed model exhibits superior performance against others, demonstrating higher prediction accuracy and stability. In practical engineering applications, the model achieves prediction accuracies exceeding 0.8, and reaches above 0.9 in partial cases, indicating satisfactory predictive effectiveness. Therefore, the proposed Optuna-optimized GBDT algorithm provides an effective solution for soil layer classification and offers valuable references for further research in this field.