Decoding Bovine Behavior: A Machine Learning Analysis of Disease and Event Detection

Abstract

The dairy industry globally serves millions, creates employment opportunities, and contributes to GDP and livelihoods, supporting numerous farmers and milk processors, and contributing to the GDP of many countries. Even with the high number of cows, a variety of illnesses such lameness, mastitis, metritis, foot and mouth disorders have a significant impact on the milk output. This demands the growth of smart methods to assist dairy producers in keeping an eye on the health, nutrition, and general well-being of their cows. To identify noteworthy bovine occurrences and illnesses, this research uses Machine Learning (ML) methods such as Extreme Gradient Boost (XGB), Naive Bayes (NB), and Perceptron. In order to uncover complex relationships and unearth undiscovered information about the characteristics and occurrences of bovine’s disease like Acidosis, Calving, Estrus, Lameness, and Mastitis. We closely scrutinize four distinct datasets. Metrics like as Area under the Curve (AUC), F1 score, accuracy, precision, recall, and Receiver Operating Characteristic (ROC) curve are used to evaluate performance. When it comes to identifying estrus events, XGB has the best detection accuracy score 92.59%, whereas XGB can detect a variety of events and disease with the highest recall 100% and the highest precision 95% and AUC of 0.962 when it comes to identifying calving, mastitis, and lameness