A Comparative Machine Learning Approach to Analyze Soil-Water Content and Its Impact on Apple Yield in Jammu and Kashmir Orchards

Abstract

The apple orchards are one of the essential part of landscape in horticultural economy of Jammu Kashmir and soil fertility and irrigation management play crucial role to sustain the productivity of the orchards. Soil characterization and understanding of how the variability in soil can affect orchard conditions is needed to enable good nutrient management decisions by farmers and for wider agricultural decision-making. The major aims of the study were:to understand the soil characteristics of apple orchards, to know the influence of irrigation water on its fertility and to evaluate the suitability of different machine learning models for analysis of soil data. Experimental Extent Of Research: The study was based on soil samples collected from different apple growing regions of Jammu and Kashmir region with the number of 5630 soil samples. The data contained the soil properties pH, Phosphorus (P), potassium (K), zinc (Zn), iron (Fe), manganese (Mn), copper (Cu), nitrogen (N), electrical conductivity (EC), organic carbon (OC) and Irrigation status. Methods followed were data preprocessing, exploratory data analysis, correlation analysis and utilizing machine learning techniques to study soil fertility patterns. Four machine learning algorithms have been tested and compared: Decision trees, Random Forests, Support Vector Machines (SVM), and Linear Regression. Soil fertility in relation to selected orchard locations analysis of soil report for two drenching compost shows that there is a lot of diversity in soil fertility from one orchard location to another.

 The soils were well distinguished into the groups based on the main four parameters; organic carbon, nitrogen, pH and potassium. A comparative analysis confirmed that the Random Forest model outperformed all other algorithms examined in terms of predictive performance. The results demonstrate the strength of machine learning in studying agricultural data and facilitate precision nutrient management and irrigation planning. This study con-tributes qualitative insights for sustainable management of apple orchards for future data-driven agricultural applications in Jammu and Kash-mir.