Hurricane intensity prediction based on time series data mining

This paper applied data mining methods to study the time-series hurricane data of 28 years (1982-2009) in Atlantic region to improve the accuracy of hurricane intensity prediction. According to the contrast test of base classifiers, IBk classifier and the LMT classifier were selected and integrated by using the Bagging method. The optimized Bagging base classifier sequence was trained using the optimal training data set after culling abnormal data. Then the hurricane intensity prediction model: the Bagging-IBk&LMT model was formed. The testing result showed that the prediction model was better than single classification model.