Scholarlink

<p>Accurate rainfall forecasting is a crucial task in meteorology and plays an essential role in agricultural planning, disaster management, water resource management, and environmental planning. However, the inherent uncertainty, vagueness, imprecision, and variability in rainfall data present significant challenges for traditional statistical and machine learning forecasting methods. In recent years, the development of Fuzzy Time Series (FTS) models has proven to be particularly effective in addressing these issues, as they are well-suited to handle uncertainty, vagueness, and imprecision in time series data. This research article employs FTS method, utilizing the frequency distribution partition of the Universe of Discourse (UoD), fuzzy logical relationship (FLRs) and fuzzy logical relationship groups (FLRGs) to represent India's annual rainfall data set. Furthermore, weighted functions are formulated based on FLRGs to enhance the accuracy of rainfall prediction. A comparative analysis between the proposed weighted FTS model and existing models demonstrates its superior performance across various statistical metrics, including Root Mean Square Error (RMSE), Absolute Forecast Error (AFER), Mean Absolute Deviation (MAD), etc. These findings highlight the model’s potential to substantially improve decision-making in sectors that are highly dependent on rainfall patterns.</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

A modified weighted model of annual rainfall forecasting using fuzzy time series techniques

  • 期刊论文

  • 发表日期：

    2025年8月3日

查看全文
  • Bhanu Pratap Singh,
  • Rinku,
  • Abhishekh,
  • Anil Kumar Nishad,
  • Vijay Kumar Patel

摘要

Accurate rainfall forecasting is a crucial task in meteorology and plays an essential role in agricultural planning, disaster management, water resource management, and environmental planning. However, the inherent uncertainty, vagueness, imprecision, and variability in rainfall data present significant challenges for traditional statistical and machine learning forecasting methods. In recent years, the development of Fuzzy Time Series (FTS) models has proven to be particularly effective in addressing these issues, as they are well-suited to handle uncertainty, vagueness, and imprecision in time series data. This research article employs FTS method, utilizing the frequency distribution partition of the Universe of Discourse (UoD), fuzzy logical relationship (FLRs) and fuzzy logical relationship groups (FLRGs) to represent India's annual rainfall data set. Furthermore, weighted functions are formulated based on FLRGs to enhance the accuracy of rainfall prediction. A comparative analysis between the proposed weighted FTS model and existing models demonstrates its superior performance across various statistical metrics, including Root Mean Square Error (RMSE), Absolute Forecast Error (AFER), Mean Absolute Deviation (MAD), etc. These findings highlight the model’s potential to substantially improve decision-making in sectors that are highly dependent on rainfall patterns.