When AI Meets Wildlife: Predicting Animal Migration from Habitat Cues

Elephant migration plays a critical role in maintaining biodiversity, yet predicting their movement remains a complex challenge influenced by environmental, human, and ecological factors. This study develops a machine learning model to forecast elephant migration between Bandipur National Park and Wayanad Wildlife Sanctuary by analyzing 34 months of historical data incorporating features like temperature, humidity, air quality index, vegetation index, and water availability index. After extensive data preprocessing, including outlier removal, feature selection, and data balancing using SMOTE, multiple machine learning algorithms were evaluated. Logistic Regression achieved the highest performance, with an accuracy of 94%, outperforming Decision Trees, Random Forests, Support Vector Machines, Naive Bayes, and Neural Networks. Exploratory data analysis revealed key environmental triggers influencing migration, such as seasonal water availability and temperature variations. Hyperparameter tuning further optimized model performance. The results demonstrate that predictive analytics can enhance conservation strategies, reduce human-elephant conflict, and support policy-making for habitat protection. Future work aims to incorporate real-time tracking and additional ecological factors to further improve model robustness and applicability in dynamic environments.