Scholarlink

The most frequent reasons for coal mining mishaps include landslides, explosions, fires, and seismic dangers. Numerous causes, including social and economic ones as well as technological and mechanical malfunctions, are typically to blame for these mishaps. By analyzing these incidents, it will be possible to pinpoint their precise causes and take steps to stop them from happening again. Regretfully, these techniques’ accuracy still needs improvement despite their advances. In this study, we used 2584 instances acquired from the UCI repository. Because of the large theft of skewed data, the entire study was broken into three sections. In the first stage, the Synthetic Minority Oversampling Technique is used to reduce the amount of skewed data. In the second stage, seven classifier models are evaluated using the following performance metrics: F1 score, accuracy, recall, precision, and training time. In the final section, we compare the suggested model’s performance. The findings depict that the suggested SMOTE-based classifier outperformed several current empirical techniques and attained high-performance metrics. The maximum accuracy obtained by this proposed method is 94.78% by the SMOTE-based ANN model.

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

Evaluating the Performance of Machine Learning Integrated SMOTE Analysis for Prediction of Risk Factors of Seismic Hazards

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Pijush Dutta,
  • Sujata Kundu,
  • Shobhandeb Paul,
  • Gour Gopal Jana,
  • Arindam Sadhu

摘要

The most frequent reasons for coal mining mishaps include landslides, explosions, fires, and seismic dangers. Numerous causes, including social and economic ones as well as technological and mechanical malfunctions, are typically to blame for these mishaps. By analyzing these incidents, it will be possible to pinpoint their precise causes and take steps to stop them from happening again. Regretfully, these techniques’ accuracy still needs improvement despite their advances. In this study, we used 2584 instances acquired from the UCI repository. Because of the large theft of skewed data, the entire study was broken into three sections. In the first stage, the Synthetic Minority Oversampling Technique is used to reduce the amount of skewed data. In the second stage, seven classifier models are evaluated using the following performance metrics: F1 score, accuracy, recall, precision, and training time. In the final section, we compare the suggested model’s performance. The findings depict that the suggested SMOTE-based classifier outperformed several current empirical techniques and attained high-performance metrics. The maximum accuracy obtained by this proposed method is 94.78% by the SMOTE-based ANN model.