Integrating random forest model and internet of things-based sensor for smart poultry farm monitoring system

Abstract

The global poultry industry has encountered growing concerns related to foodborne illnesses, misuse of antibiotics, and environmental impacts. To tackle these issues, this study aims to develop an intelligent poultry farm with real-time environmental monitoring and predictive models. The primary objective is to combine a machine learning-based prediction model with internet of things (IoT) devices to gather and analyze environmental data, such as temperature, humidity, and ammonia levels, to forecast the conditions within poultry houses. These sensor data and additional information, such as feed consumption, water consumption, poultry weight, capacity, and poultry house dimensions will serve as inputs for supervised machine learning models. Among these models, the proposed random forest (RF) model, when augmented with timestamp features, achieves the highest accuracy rate of 96.665%, surpassing other models such as logistic regression (LR), k-nearest neighbor (KNN), decision tree (DT), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), support vector machine (SVM), and multi-layer perceptron (MLP) in identifying poultry house conditions. Additionally, this study demonstrates how the trained model can be effectively applied in a web-based monitoring system, delivering real-time data to farmers for well-informed decision-making and ultimately enhancing productivity in smart poultry farming.