Research on Pest Propagation Based on Auto-Regressive and Moving Average Model Algorithm

Abstract

The presence of Vespa mandarinia can have a potentially serious impact on local bee populations and should be removed as soon as possible. In order to eradicate the Vespa mandarinia, we present several guidelines and strategies to help the State of Washington to allocate and utilize the limited resources efficiently.We describe our process in terms of CUU, a novel framework for Model Creation, Use and Update. On the basis of the ecological content of pests and the positive ID, negative ID and unprocessed data from the data table, we concluded that the spread of the pest changed over time. Afterwords, we infer that the range of hornet is small from the problem. So we utilize Auto-Regressive and Moving Average Model(ARMA) model as the time series prediction method and residual analysis to achieve the prediction accuracy in the paper. Later on, we classified the data utilizing K-nearest neighbor algorithm(KNN), and obtain that the unprocessed data were all Negative ID. Using the Positive ID data again, we select one of the points, calculate the average distance from the remaining 13 points to that point and calculate the probability of the presence of pests around that point, finally achieve the probability of mistaken classification: The smaller the average distance and the greater the probability of the presence of pests, the smaller the probability of the misclassification of data within 30 km of this area. Furthermore, to investigate the government’s desire to optimize resource allocation, entropy weight method and Technique for Order Preference by Similarity to an Ideal Solution(TOPSIS) method are proposed to score the locations where pests have been confirmed to appear. Finally, these locations are ranked by the score, where the harmful organisms are more likely to appear around the area when the score of this area is higher.