An Optimized Neural Network Model to Predict Milk Quality

Milk quality prediction is essential in the dairy industry to optimize production processes, ensure regulatory compliance, and maintain consumer trust. By leveraging predictive models that consider factors like fat and protein content, somatic cell count, and bacterial load, farmers can make informed decisions about feed composition, animal health, and milking practices. For dairy processors, accurate predictions aid supply chain management, production scheduling, and product safety. In this research, we proposed a model using neural networks to predict milk quality based on particulars like color, fat content, and pH. Our priority is highly given to predicting quality using these features. We created an optimized neural network by reducing parameters. Our model delivered an accuracy of 0.95. However, we understand that accuracy alone does not encompass performance evaluation. Thus, we delved deeper, considering metrics like Fl-score, precision, and recall. In addition to these assessments, we conducted a robust validation using the bootstrap method. This comprehensive validation revealed that our model performed well and avoided overfitting and underfitting, solidifying its reliability and robustness in predicting milk quality.