Scholarlink

<p>This study introduces an Enhanced Hyper-Parameter Configuration (EHPC) framework proposed to optimize hyperparameters of tree-based estimators for precipitation prediction models. The framework integrates Enhanced Bayesian optimization (EBO) approach, the Expected Improvement with Probability of Improvement acquisition function, and the simultaneous seeking strategy to enhance both efficiency and accuracy. The non-linear patterns and complex relationships within the meteorological features are captured using tree-based estimators such as Random forest and Gradient boosting. By modeling prediction performance, the EBO efficiently explores the hyperparameter space. EBO effectively explores the hyperparameter space using a surrogate model that iteratively selects promising configurations, minimizing the cost of evaluating the objective function. The EIPI function balances exploration and exploitation by combining the strengths of Probability of Improvement and Expected Improvement. The integration of EBO within the EHPC framework ensures that hyperparameters are chosen based on both predictive performance and computational efficiency. Overall, integrating EBO with tree-based estimators significantly improves the accuracy and reliability of precipitation prediction. The proposed strategy outperforms other existing approaches with a high prediction accuracy of 98.99% with an execution time of 28.99&#xa0;s. This framework significantly improves prediction accuracy while reducing optimization time, making it a promising methodology for efficient precipitation prediction tasks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Efficient Hyperparameter Tuning for Precise Precipitation Prediction Using EHPC Framework

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Nivethitha K,
  • Lokeshkumar R

摘要

This study introduces an Enhanced Hyper-Parameter Configuration (EHPC) framework proposed to optimize hyperparameters of tree-based estimators for precipitation prediction models. The framework integrates Enhanced Bayesian optimization (EBO) approach, the Expected Improvement with Probability of Improvement acquisition function, and the simultaneous seeking strategy to enhance both efficiency and accuracy. The non-linear patterns and complex relationships within the meteorological features are captured using tree-based estimators such as Random forest and Gradient boosting. By modeling prediction performance, the EBO efficiently explores the hyperparameter space. EBO effectively explores the hyperparameter space using a surrogate model that iteratively selects promising configurations, minimizing the cost of evaluating the objective function. The EIPI function balances exploration and exploitation by combining the strengths of Probability of Improvement and Expected Improvement. The integration of EBO within the EHPC framework ensures that hyperparameters are chosen based on both predictive performance and computational efficiency. Overall, integrating EBO with tree-based estimators significantly improves the accuracy and reliability of precipitation prediction. The proposed strategy outperforms other existing approaches with a high prediction accuracy of 98.99% with an execution time of 28.99 s. This framework significantly improves prediction accuracy while reducing optimization time, making it a promising methodology for efficient precipitation prediction tasks.