Scholarlink

<p>Agriculture plays a crucial role in India’s economic development, with crop productivity influenced by population growth and climate change. Data mining provides valuable insights into patterns and models for vast data volumes, and agricultural productivity growth is crucial for poverty alleviation. However, inadequate technical guidance can lead to varying crop yields. Existing techniques face challenges due to complexity and uncertainty, resource shortages, poor real-time responsiveness, and low detection accuracy. To address the aforementioned difficulties, an effective deep residual fuzzy encoder approach is developed for forecasting crop yield in agricultural land. Crop yield data are collected and utilized as an input data set. Initially the collected input datasets are pre-processed using random drop imputation to replace the missing values, feature-wise normalization is used to standardize the input data and the inter quartile range is applied for outliers detections on the data. Next, the pre-processed data are given to feature selection using the ReliefF (RF) technique. The number of neighbours in the RF is optimally chosen by the barnacles mating optimization (BMO) approach. Kho-Kho Optimization (KKO) approach is utilized to select the hyper-parameters such as membership functions, and number of fuzzy set of the deep residual fuzzy encoder approach in an optimal manner to forecast crop yield. Based on simulation findings, the proposed method achieves 95.6% accuracy, 10% FDR and 89.85% F1_Score values. Thus, this proposed approach is the best choice for forecasting crop yield on agricultural land.</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

Agriculture crop yield forecasting using relief based feature selection and KHO-KHO optimization approach for deep residual fuzzy encoders

  • 期刊论文

  • 发表日期：

    2025年10月30日

查看全文
  • Anisa B. Shikalgar,
  • Tahseen A. Mulla,
  • Nasim Husen Sayyad,
  • Karishma B. Shambharkar,
  • Sayali Ganesh Sanmukh

摘要

Agriculture plays a crucial role in India’s economic development, with crop productivity influenced by population growth and climate change. Data mining provides valuable insights into patterns and models for vast data volumes, and agricultural productivity growth is crucial for poverty alleviation. However, inadequate technical guidance can lead to varying crop yields. Existing techniques face challenges due to complexity and uncertainty, resource shortages, poor real-time responsiveness, and low detection accuracy. To address the aforementioned difficulties, an effective deep residual fuzzy encoder approach is developed for forecasting crop yield in agricultural land. Crop yield data are collected and utilized as an input data set. Initially the collected input datasets are pre-processed using random drop imputation to replace the missing values, feature-wise normalization is used to standardize the input data and the inter quartile range is applied for outliers detections on the data. Next, the pre-processed data are given to feature selection using the ReliefF (RF) technique. The number of neighbours in the RF is optimally chosen by the barnacles mating optimization (BMO) approach. Kho-Kho Optimization (KKO) approach is utilized to select the hyper-parameters such as membership functions, and number of fuzzy set of the deep residual fuzzy encoder approach in an optimal manner to forecast crop yield. Based on simulation findings, the proposed method achieves 95.6% accuracy, 10% FDR and 89.85% F1_Score values. Thus, this proposed approach is the best choice for forecasting crop yield on agricultural land.