Scholarlink

This paper explores the analysis and automatic prediction of macro- and micronutrients using machine learning approaches to comprehend soil health. Through this research we are exploring the analytical capabilities of Machine learning algorithm for imputing missing values specifically for macro and micronutrients of soil. The various types of machine learning algorithms used are Decision Tree, Liner Regression and Ensemble techniques like Random Forest, XGBoost and ADABoost. A total of 25 villages (with varying chemical and texture properties) from Ambegaon Taluka, Pune District, Maharashtra were sampled at a depth of 0–30 cm for laboratory analysis. It is composed of the macronutrients potassium (K), phosphorus (P), and nitrogen (N). However, other micronutrients such as Zinc (Zn), Iron (Fe), Copper (Cu), Manganese (Mn), Sulphur (S), Boron (B), Organic Carbon (OC), Electrical Conductivity (EC) and pH are also considered. R2 values using Random Forest (RF) for pH(0.9221), OC (0.9992) N (0.99936), P (0.50291), K (0.9561) and Zn (0.9694); using Decision Tree EC (0.9486), K (0.9651) and B (0.9999) and using XGBoost N (0.9993), S (0.995), Fe (0.9662), Cu (0.9988) and Mn (0.9907). It can be concluded with this research that ML algorithms can be utilized for imputing missing values of soil nutrients. They can be deployed as secondary measures for reducing the cost and efforts of repeating primary tests for soil nutrient testing in case of unacceptable values calculation through these methods. Also, a longitudinal study considering crop cultivation and other environmental aspects can help in formalizing a standard for this domain.

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

Soil Health Analysis and Automatic Prediction of Macro and Micronutrient Using Machine Learning Algorithms

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Varsha Atul Shukre,
  • Surabhi Thatte,
  • Aishwarya Sameer Karandikar,
  • Chaitali Pradeep Kannurkar,
  • Amey Madhukar Patil,
  • Bhavesh Sudhir Jagtap

摘要

This paper explores the analysis and automatic prediction of macro- and micronutrients using machine learning approaches to comprehend soil health. Through this research we are exploring the analytical capabilities of Machine learning algorithm for imputing missing values specifically for macro and micronutrients of soil. The various types of machine learning algorithms used are Decision Tree, Liner Regression and Ensemble techniques like Random Forest, XGBoost and ADABoost. A total of 25 villages (with varying chemical and texture properties) from Ambegaon Taluka, Pune District, Maharashtra were sampled at a depth of 0–30 cm for laboratory analysis. It is composed of the macronutrients potassium (K), phosphorus (P), and nitrogen (N). However, other micronutrients such as Zinc (Zn), Iron (Fe), Copper (Cu), Manganese (Mn), Sulphur (S), Boron (B), Organic Carbon (OC), Electrical Conductivity (EC) and pH are also considered. R2 values using Random Forest (RF) for pH(0.9221), OC (0.9992) N (0.99936), P (0.50291), K (0.9561) and Zn (0.9694); using Decision Tree EC (0.9486), K (0.9651) and B (0.9999) and using XGBoost N (0.9993), S (0.995), Fe (0.9662), Cu (0.9988) and Mn (0.9907). It can be concluded with this research that ML algorithms can be utilized for imputing missing values of soil nutrients. They can be deployed as secondary measures for reducing the cost and efforts of repeating primary tests for soil nutrient testing in case of unacceptable values calculation through these methods. Also, a longitudinal study considering crop cultivation and other environmental aspects can help in formalizing a standard for this domain.