PREDICTING LAND SUITABILITY FOR CEREAL CROPS USING MACHINE LEARNING APPROACHES

Advisor: Mr.Kamal Mohammed(Ass.Prof) Assessing land suitability is a critical agricultural practice aimed at improving crop productivity. Nevertheless, ensuring food security remains a significant challenge, especially in developing nations like Ethiopia, where population growth intensifies the demand for agricultural output. Traditional land suitability assessment methods are often labor-intensive, time-consuming, costly, and lack the precision and speed needed for timely decision-making. To overcome these challenges, this research utilizes machine learning methods to forecast land suitability for five key cereal crops cultivated in Ethiopia namely Wheat, Barley, Maize, Sorghum, and Teff based on datasets acquired from the Engineering Corporation of Oromia. The dataset underwent extensive preprocessing to prepare it for model training. Three machine learning algorithms Random Forest (RF), Gradient Boosting (GB), and K-Nearest Neighbors (KNN) were employed alongside feature selection methods including Univariate Feature Selection (UFS), Recursive Feature Elimination with Cross-Validation (RFECV), and Sequential Forward Selection (SFS). Model performance was optimized using randomized search with cross-validation for hyperparameter tuning. Each model was evaluated under stratified 10-fold cross-validation using performance metrics such as accuracy, precision, recall, F1-score, and confusion matrix. The findings indicated that the Gradient Boosting algorithm, when integrated with Sequential Forward Selection, delivered superior performance achieving an accuracy of 94.41%, a precision of 94.37%, a recall of 94.34%, and an F1-score of 94.35%. Therefore, the GB model paired with SFS is recommended as the most effective approach for accurately and efficiently predicting land suitability for cereal crop cultivation. In addition to developing and evaluating machine learning models, this study suggests that future research could explore deep learning approaches with larger datasets to further enhance prediction accuracy.
Keywords: Land Suitability, Cereal Crops, Machine Learning, Gradient Boosting.