Scholarlink

<p>This study presents a comprehensive quantitative analysis and machine learning-based prediction of rainfall patterns across six major Indian cities: Hyderabad, Delhi, Mumbai, Chennai, Kolkata, and Bangalore. The research distinguishes statistical analysis techniques, such as principal component analysis (PCA) – with an explained variance ratio of 86.7% for PCA1 and 13.3% for PCA2 – and seasonal decomposition, revealing trend means ranging from 1.7 mm in Delhi to 4.9 mm in Kolkata, from machine learning-based predictions utilizing models like XGBoost (accuracy: 76.5%), random forest (accuracy: 76.6%), and gradient boosting (accuracy: 76.6%). The selection of these models is guided by their proven efficiency in handling medium-sized structured datasets, although future work may explore deep learning and hybrid approaches to enhance performance. Seasonal differences are highlighted, such as monthly rainfall peaks of 12.1 mm in Chennai (November), 20.2 mm in Kolkata (July), and 18.5 mm in Mumbai (July). The findings underscore practical implications for real-world applications, including early warning systems and climate adaptation strategies. Validation against historical records strengthens the reliability of trends, providing valuable insights for urban planning and disaster management.</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

Unravelling rainfall trends: A comprehensive study of urban precipitation using innovative statistical and machine learning techniques in major cities of India

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • V Guhan,
  • A Dharma Raju,
  • A Sravani,
  • K Naga Ratna

摘要

This study presents a comprehensive quantitative analysis and machine learning-based prediction of rainfall patterns across six major Indian cities: Hyderabad, Delhi, Mumbai, Chennai, Kolkata, and Bangalore. The research distinguishes statistical analysis techniques, such as principal component analysis (PCA) – with an explained variance ratio of 86.7% for PCA1 and 13.3% for PCA2 – and seasonal decomposition, revealing trend means ranging from 1.7 mm in Delhi to 4.9 mm in Kolkata, from machine learning-based predictions utilizing models like XGBoost (accuracy: 76.5%), random forest (accuracy: 76.6%), and gradient boosting (accuracy: 76.6%). The selection of these models is guided by their proven efficiency in handling medium-sized structured datasets, although future work may explore deep learning and hybrid approaches to enhance performance. Seasonal differences are highlighted, such as monthly rainfall peaks of 12.1 mm in Chennai (November), 20.2 mm in Kolkata (July), and 18.5 mm in Mumbai (July). The findings underscore practical implications for real-world applications, including early warning systems and climate adaptation strategies. Validation against historical records strengthens the reliability of trends, providing valuable insights for urban planning and disaster management.