Scholarlink

<p>Crop yield is influenced by a&#xa0;variety of interdependent traits, rendering yield prediction a&#xa0;statistically challenging endeavor. Worldwide, kiwifruit is considered unique because of its nutrient contents and good taste. This study endeavors to predict the yield of kiwifruit by utilizing morphological characteristics. The artificial neural network (ANN) surpasses other models by attaining the lowest mean absolute percentage error (MAPE) of 4.655, which signifies the highest level of prediction accuracy. This is followed by the support vector machine (SVM), which has a&#xa0;moderate MAPE of 6.692. Additionally, ANN exhibits the lowest root mean squared error (RMSE) at 0.812 and a&#xa0;normalized RMSE (nRMSE) of 6.266, effectively reducing both absolute and normalized errors. Its dominance is further supported by the highest R‑squared value of 0.882 and a&#xa0;Pearson correlation coefficient of r = 0.939, underscoring its exceptional capability to account for variance and its strong linear relationship with the observed data. The current research enhances our comprehension of the intricate connections between crop yield and morphological characteristics.</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

Predicting Kiwifruit Yield Using Machine Learning: Insights from Morphological Characteristics

  • 期刊论文

  • 发表日期：

    2025年9月2日

查看全文
  • Sudip Kumar Dutta,
  • Tanuj Misra,
  • Samir Barman,
  • Shailendra Kumar,
  • Mrinmoy Ray

摘要

Crop yield is influenced by a variety of interdependent traits, rendering yield prediction a statistically challenging endeavor. Worldwide, kiwifruit is considered unique because of its nutrient contents and good taste. This study endeavors to predict the yield of kiwifruit by utilizing morphological characteristics. The artificial neural network (ANN) surpasses other models by attaining the lowest mean absolute percentage error (MAPE) of 4.655, which signifies the highest level of prediction accuracy. This is followed by the support vector machine (SVM), which has a moderate MAPE of 6.692. Additionally, ANN exhibits the lowest root mean squared error (RMSE) at 0.812 and a normalized RMSE (nRMSE) of 6.266, effectively reducing both absolute and normalized errors. Its dominance is further supported by the highest R‑squared value of 0.882 and a Pearson correlation coefficient of r = 0.939, underscoring its exceptional capability to account for variance and its strong linear relationship with the observed data. The current research enhances our comprehension of the intricate connections between crop yield and morphological characteristics.