Modeling for Soil Parameters Based on Multinomial Logistic Regression

This study aimed to investigate the pH associated with micronutrients in soil samples. The characterization of the soil involved analyzing various factors, including pH, Sulfur, Zinc, Iron, Copper, Manganese, and Boron. A total of 500 soil samples were collected and categorized into four pH ranges: moderately acidic (5.1-6.0%), slightly acidic (6.1-6.5%), neutral (6.6-7.5%), and slightly alkaline (7.6-8.5%). The pH levels were neutral, indicating an ideal condition for maximum availability of primary nutrients essential for plant growth. The collected data was subjected to multinomial logistic regression and multivariate linear regression analyses. A formula was derived to determine the soil micronutrients based on the corresponding pH levels. The statistical tests of significance using linear regression indicated significant differences (P>0.05) between the pH values of the soil samples for Sulfur, Manganese, and Boron. Additionally, correlation analysis explored the relationships among different soil parameters. The likelihood ratio test supported a relationship between pH and micronutrient levels. The findings demonstrated that pH levels can be a predictive indicator of micronutrient performance in soil. Model evaluation, including the goodness of fit tests and pseudo-R-squares, accounted for 17% of the overall assessment. Moreover, the Multinomial logistic regression analysis achieved a classification accuracy of 64% for predicting pH levels. These findings have important implications for effective soil management strategies, aiding in optimizing nutrient availability for plant growth.