Multi Classification Unbalance Data Model Based On Bagging-CART

With the development of technology and the application of a large number of high-tech sensors,civil aviation has entered the era of data.Aircraft operations and maintenance will generate a large amount of maintenance-related data,which will generate huge demand for the full use of data resources.Among a large number of data mining algorithms,Bagging-CART has been widely used due to its simplicity,ease of use,and ease of parallel execution.At the same time,the data in reality is often unbalanced,as in the failure data,some failures are less than others,but the consequences are more significant.The Bagging-CART algorithm is proposed under the conditions of data balance.In the face of unbalanced data,it is biased to identify a few kinds of faults as multi-data faults,resulting in poor practicality.This research focuses on the data mining method of civil aviation maintenance data multi-class unbalanced data characteristics,thus effectively providing key information for maintenance decision.By proposing a comprehensive processing plan,we start with the data preprocessing and the decision-making methods of the Bagging-CART classifier.In terms of data,a Bagging-based balanced data preprocessing method is proposed,and the sampling process of the algorithm is controlled on the basis of the insufficiency of Bagging.Finally,the data is converted into a balanced data group without changing the original data.In terms of algorithm,an optimization method is proposed for the decision-making process of Bagging-CART classifier.By introducing a weight-based minimum distance model and improving it,the relationship between sample data and test data is introduced into the classification process..Through the comprehensive application of the two methods,this model has a good ability to deal with multi-class unbalanced problems,laying a certain foundation for the practicality of the model.Finally,a data mining system is designed based on the algorithm of this paper,which can be used for data mining of civil aviation maintenance data,and lays a foundation for the application research of data analysis model.