Scholarlink

Crop yield prediction for Indian states is a significant topic in agriculture, utilizing machine learning and statistical models to forecast production. It helps farmers, policymakers, and researchers optimize resources and improve food security. This study aims to analyse and predict crop yield across various states of India using machine learning approaches. This study comprises agricultural data for a variety of crops (55 different crops) cultivated across different states (30 states) in India, spanning the years 1997 to 2020. It includes key features essential for crop yield prediction, such as crop type, crop year, cropping season, state-wise cultivation areas, production volumes, annual rainfall, fertilizer and pesticide usage, and calculated crop yields. This study aims to develop feature engineering based machine learning models to predict crop yield of top ten crops of Indian States. This study has tested and compared the performance of ML models (DT, RF, SVR, XGB and DNN) across new set of features (derived features) on top ten different crops. Among all the models, XGBoost and DNN outperform others in capturing nonlinear relationships, especially for high-yielding crops like Sugarcane and Wheat.

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

Crop Yield Prediction of Indian States Based on Machine Learning and Feature Engineering

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Smita Agrawal,
  • Rajeev Kumar,
  • Tadiwa Elisha,
  • Abu Bakar Abdul Hamid

摘要

Crop yield prediction for Indian states is a significant topic in agriculture, utilizing machine learning and statistical models to forecast production. It helps farmers, policymakers, and researchers optimize resources and improve food security. This study aims to analyse and predict crop yield across various states of India using machine learning approaches. This study comprises agricultural data for a variety of crops (55 different crops) cultivated across different states (30 states) in India, spanning the years 1997 to 2020. It includes key features essential for crop yield prediction, such as crop type, crop year, cropping season, state-wise cultivation areas, production volumes, annual rainfall, fertilizer and pesticide usage, and calculated crop yields. This study aims to develop feature engineering based machine learning models to predict crop yield of top ten crops of Indian States. This study has tested and compared the performance of ML models (DT, RF, SVR, XGB and DNN) across new set of features (derived features) on top ten different crops. Among all the models, XGBoost and DNN outperform others in capturing nonlinear relationships, especially for high-yielding crops like Sugarcane and Wheat.