Scholarlink

<p>The integration of Internet of Things (IoT) concepts and data mining has significantly transformed industries sensitive to environmental variability, such as agriculture, retail, and transportation. While machine learning techniques have been widely applied to weather and economic analysis, existing studies often rely on historical climate data, focus on single domains, and lack interpretability for decision-making. This paper addresses these gaps by presenting an IoT-inspired analytical framework that employs the Random Forest algorithm to model the impact of weather conditions on economic activities. The methodology begins by leveraging publicly available benchmark datasets (ECA&amp;D and Kaggle) to simulate real-time IoT conditions. While the current study applies these datasets, the proposed framework is designed for seamless integration with live IoT sensor networks in future deployments, aligning the weather data with sector-specific economic indicators. Data preprocessing, feature engineering, and feature selection (utilizing Pearson correlation, RFE, and mutual information) are employed to extract the most relevant weather variables, including temperature, precipitation, and humidity. The Random Forest model captures complex, non-linear interactions between weather variables and economic performance across multiple sectors. Comparative experiments demonstrate that the proposed framework achieves high predictive performance, with an accuracy of 94.56%, precision of 92.83%, recall of 93.45%, F1 score of 92.75%, and cross-validation accuracy of 91.75%, outperforming traditional machine learning and advanced ensemble methods, such as Decision Trees, SVM, XGBoost, and LightGBM. Furthermore, interpretability methods (SHAP values, feature importance) provide transparent insights for policymakers and industry stakeholders. Although validated using historical datasets, the framework is designed for future&#xa0;deployment with real-time IoT sensor networks, enabling adaptive and robust economic&#xa0;forecasting under changing climate conditions.</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

IoT-Based Data Mining Analysis of the Impact of Weather Data on Economic Activities Using Random Forest Algorithms

  • 期刊论文

  • 发表日期：

    2025年12月22日

查看全文
  • Lili Lyu,
  • Fang Xiao,
  • Chunxiang Leng,
  • Dan Yu,
  • Yixi Fan,
  • Ping Li,
  • Binbing Zhang

摘要

The integration of Internet of Things (IoT) concepts and data mining has significantly transformed industries sensitive to environmental variability, such as agriculture, retail, and transportation. While machine learning techniques have been widely applied to weather and economic analysis, existing studies often rely on historical climate data, focus on single domains, and lack interpretability for decision-making. This paper addresses these gaps by presenting an IoT-inspired analytical framework that employs the Random Forest algorithm to model the impact of weather conditions on economic activities. The methodology begins by leveraging publicly available benchmark datasets (ECA&D and Kaggle) to simulate real-time IoT conditions. While the current study applies these datasets, the proposed framework is designed for seamless integration with live IoT sensor networks in future deployments, aligning the weather data with sector-specific economic indicators. Data preprocessing, feature engineering, and feature selection (utilizing Pearson correlation, RFE, and mutual information) are employed to extract the most relevant weather variables, including temperature, precipitation, and humidity. The Random Forest model captures complex, non-linear interactions between weather variables and economic performance across multiple sectors. Comparative experiments demonstrate that the proposed framework achieves high predictive performance, with an accuracy of 94.56%, precision of 92.83%, recall of 93.45%, F1 score of 92.75%, and cross-validation accuracy of 91.75%, outperforming traditional machine learning and advanced ensemble methods, such as Decision Trees, SVM, XGBoost, and LightGBM. Furthermore, interpretability methods (SHAP values, feature importance) provide transparent insights for policymakers and industry stakeholders. Although validated using historical datasets, the framework is designed for future deployment with real-time IoT sensor networks, enabling adaptive and robust economic forecasting under changing climate conditions.