Abstract

Late identification of swine diseases results in significant economic loss associated with swine farming. Therefore, it is important to identify predictors of swine health condition. This study presents assessments of swine health using indicators of symptoms characteristics of swine based on machine learning algorithms. Three machine learning classification algorithms (Decision Tree (DT) and Logistic Regression and Support Vector Machine (SVM)) was employed to develop an expert system to capture the knowledge of an expert (endocrinologists) and made it available through a software interface (GUI) for detection, diagnosis and treatment of swine diseases. Comparisons of these algorithms were carried out based on performance factors which include: classification accuracy, precision, sensitivity and specificity. The result showed that Decision Tree produced the highest accuracy of 97.5% as compared to Support Vector Machine which produced 88.8% accuracy, and Logistic Regression which produced 75% accuracy. Hence, Decision Tree is identified as the best performing algorithm for detecting, diagnosing and treatment of swine diseases using a small set of simple identifiable symptoms and clinical measurements at an early stage of animal disease development.