Fisher Linear Discriminant Modeling for Crop Classifications Based on Soil Attributes

Abstract

This study employed multivariate statistical techniques to analyze soil nutrient data for crop classification, focusing on the "Potato" and "Raagi" crops. The analysis revealed highly significant differences in soil nutrient profiles between these crop types, with specific soil nutrients exhibiting substantial variability. The Fisher Linear Discriminant Analysis demonstrated exceptional discriminative power, achieving perfect crop separation. The confusion matrix indicated high classification accuracy, with "Potato" reaching 100% accuracy and "Ragi" at 96.15%. The ROC value of 0.992 further validated the model's effectiveness in crop discrimination. These findings highlight the utility of multivariate statistical approaches for crop classification and selection based on soil nutrient characteristics.