Activity Monitoring and Detection of Rabbit Livestock Using Internet of Things (IoT) with Classification Tree (CT) Method

This research addresses the challenges of real-time rabbit activity monitoring in farm management by implementing an IoT-based system with a Classification Tree (CT) method. The system uses an MPU6050 motion sensor integrated with a NodeMCU ESP32 microcontroller to collect acceleration and gyroscope data, which is transmitted to a MySQL database for analysis. The CT algorithm classifies rabbit activities into three categories: sleeping, eating, and moving. A balanced dataset of 1,768 samples ensured unbiased training, with the model achieving an accuracy of 72.13%. While unbiased, the decision tree structure introduces a risk of overfitting, which may affect classification performance. A comparison with the Random Forest algorithm, which achieved an accuracy of 78.34%, indicates the potential benefits of exploring alternative classification methods. Additional features, such as environmental sensors or improved preprocessing techniques, could refine the classification process. This research demonstrates the use of CT for IoT-based rabbit activity monitoring and provides a structured approach to analyzing motion data in livestock management. Future studies could focus on optimizing the model, expanding its application to other farm animals, and integrating real-time decision support for farmers.