Scholarlink

<p>As the global population continues to grow, the demand for increased food production has become a pressing concern, necessitating closer examination of agricultural challenges. Fortunately, modern agricultural practices offer efficient solutions to address these issues. These practices encompass various aspects, such as managing irrigation, enhancing crop yields, controlling pests and weeds, and optimizing fertilizer application. Enhancing crop yield is a pivotal component of effective agriculture management, and it relies heavily on factors such as soil quality, water availability, and prevailing climate conditions. Cutting-edge technologies such as the Internet of Things (IoT), Data Mining, Cloud Computing, and Machine Learning (ML) have emerged as instrumental tools in agriculture. This study attempts to assess the efficacy of various ML regression techniques, including random forest, gradient boost, AdaBoost, decision tree, support vector regression, lasso, ridge, elastic net, and multilinear regression, in predicting sugarcane yields. The evaluation involved the use of an extensive well-curated dataset encompassing soil parameters collected over a span of approximately 15 years over a study area in Maharashtra, India, through extensive field visits. The collected data considered different regions of crops such as plains, sub-mountains, and droughts prone to have more diversity in model training and learnability. By scrutinizing the relationships between sugarcane yield and a diverse range of climate and soil parameters, this study aimed to find the best ML algorithm for predicting sugarcane crop yield in the Indian context. The experimental findings show that gradient boost regression (GBR) outperforms over other models with an impressive average adjusted R<sup>2</sup> score of 72.11 across all three regions and a minimal average loss of 11.7. The obtained results are comparable against existing studies that consider soil and climate parameters. In conclusion, this study underscores the potential benefits of early sugarcane yield prediction (YP) in enabling farmers to enhance their crop production and consequently improve their socioeconomic status.</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

Sugarcane Crop Yield Prediction Using Supervised Machine Learning Techniques

  • 期刊论文

  • 发表日期：

    2026年1月7日

查看全文
  • Vijayatai Hukare,
  • Vidya Kumbhar,
  • Sahil K. Shah

摘要

As the global population continues to grow, the demand for increased food production has become a pressing concern, necessitating closer examination of agricultural challenges. Fortunately, modern agricultural practices offer efficient solutions to address these issues. These practices encompass various aspects, such as managing irrigation, enhancing crop yields, controlling pests and weeds, and optimizing fertilizer application. Enhancing crop yield is a pivotal component of effective agriculture management, and it relies heavily on factors such as soil quality, water availability, and prevailing climate conditions. Cutting-edge technologies such as the Internet of Things (IoT), Data Mining, Cloud Computing, and Machine Learning (ML) have emerged as instrumental tools in agriculture. This study attempts to assess the efficacy of various ML regression techniques, including random forest, gradient boost, AdaBoost, decision tree, support vector regression, lasso, ridge, elastic net, and multilinear regression, in predicting sugarcane yields. The evaluation involved the use of an extensive well-curated dataset encompassing soil parameters collected over a span of approximately 15 years over a study area in Maharashtra, India, through extensive field visits. The collected data considered different regions of crops such as plains, sub-mountains, and droughts prone to have more diversity in model training and learnability. By scrutinizing the relationships between sugarcane yield and a diverse range of climate and soil parameters, this study aimed to find the best ML algorithm for predicting sugarcane crop yield in the Indian context. The experimental findings show that gradient boost regression (GBR) outperforms over other models with an impressive average adjusted R2 score of 72.11 across all three regions and a minimal average loss of 11.7. The obtained results are comparable against existing studies that consider soil and climate parameters. In conclusion, this study underscores the potential benefits of early sugarcane yield prediction (YP) in enabling farmers to enhance their crop production and consequently improve their socioeconomic status.