Predicting Aflatoxin Contamination in Crops Using Machine Learning Algorithms

Aflatoxin is a common contaminant in grains and cereals, creating serious health hazards for consumers and economic challenges for producers. Reliable prediction of whether its levels surpass safe thresholds using weather data, soil properties, and farming practices is essential for guiding informed decisions. In this study, four machine learning algorithms: Gaussian Process Classification (GPC), Support Vector Machine (SVM), Random Forest Classifier (RFC), and K-Nearest Neighbours (KNN) were applied to predict aflatoxin contamination in maize and groundnuts. GPC outperformed other models by predicting correctly 92 out 100 groundnuts samples (92%) and 93 out of 100 maize samples (93%). The findings indicate that humidity and rainfall are stronger predictors of aflatoxin contamination compared to temperature or soil p H. This work represents an important step toward applying machine learning techniques for aflatoxin prediction in crops. Nevertheless, the study is primarily simulation-based, serving to highlight the potential of machine learning models when applied to available datasets.