Prediction of the appearance of monkeys based on weather data and wireless sensing network

Recently agricultural damage caused by monkeys is a critical problem in Japan. We propose a system that predicts what date and time monkeys approach into farmland. In order to make predictions, we had monkeys wear collars with transmitters and collected the trend data of the distances of monkeys by the receivers in this system. The time of appearing monkeys, and the weather data are taken for the estimation. We adopted both of Bayesian estimation as a linear method and Support Vector Machine as a non-linear method. As a result, the Bayesian estimation had little effect, but the Support Vector Machine outperformed the Bayesian estimation, so the effectiveness of the predicting system of monkeys by the weather data was confirmed.