This paper aims to predict soil properties through the use of machine learning techniques, providing farmers with valuable information for efficient farming and increased crop yields. The key soil properties of interest include Calcium, Phosphorus, pH, Soil Organic Carbon, and Sand, all of which have a significant impact on crop production. The study compares four popular machine learning models, including multiple linear regression, random forest regression, support vector machine, and gradient boosting, using the Africa Soil Property Prediction dataset for evaluation. Results show that gradient boosting is the most effective model in terms of coefficient of determination, accurately predicting all soil properties except for phosphorus. This predictive approach is highly beneficial for farmers to optimize their farming strategies and resource usage by obtaining critical information about soil properties in their specific location.

Machine learning, soil quality, mutation, Artificial Intelligence.