Scholarlink

<p>One of the most important features that will enhance future road safety in India, and which will be focused on predicting accidents is that road accidents usually lead to the deaths of young people. Unique causes that cause primary accident need to be predicted for forecasting of accidents in an area. For this purpose, NH-57 road is selected and accidents caused in this highway from 2020 to 2022 has been taken. Along with this traffic and geometric parameters of this road is taken, then all the data are combined to form a single data. This data is pre-processed for missing value removal and redundant value removal. This pre-processed data is then given to Pearson correlation to find the correlated data. High intensity data is then clustered using Si-FINCH clustering, with high clustering efficiency, this clustering resulted in fatal, injuries and deaths. To improve this clustering sensitivity analysis were performed, features were extracted and then given to clustering for improvement. Clustering results were then given to Kernel Density map for mapping block spot areas. Using EDQ classifier, black spots are classified with high accuracy of about 92%. Overall, the performance of the proposed model is far better than previous algorithms.</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

Forecasting accidents using Si-FINCH and EDQ classifier On NH-57 highway roads in Bihar

  • 期刊论文

  • 发表日期：

    2025年8月8日

查看全文
  • Vinod Kumar,
  • Sanjeev Kumar Suman

摘要

One of the most important features that will enhance future road safety in India, and which will be focused on predicting accidents is that road accidents usually lead to the deaths of young people. Unique causes that cause primary accident need to be predicted for forecasting of accidents in an area. For this purpose, NH-57 road is selected and accidents caused in this highway from 2020 to 2022 has been taken. Along with this traffic and geometric parameters of this road is taken, then all the data are combined to form a single data. This data is pre-processed for missing value removal and redundant value removal. This pre-processed data is then given to Pearson correlation to find the correlated data. High intensity data is then clustered using Si-FINCH clustering, with high clustering efficiency, this clustering resulted in fatal, injuries and deaths. To improve this clustering sensitivity analysis were performed, features were extracted and then given to clustering for improvement. Clustering results were then given to Kernel Density map for mapping block spot areas. Using EDQ classifier, black spots are classified with high accuracy of about 92%. Overall, the performance of the proposed model is far better than previous algorithms.