A New Classification Model For Imbalanced Classification

Imbalanced multi classification problem is a hot topic in the research of classification,which has a wide range of applications,such as anomaly detection,disease diagnosis and so on.At the same time,unbalanced multi classification problem is also a difficult problem in classification.Firstly,the sample distribution is not uniform,and the number of samples is different between different categories;Secondly,it is a multi-classification problem.In this paper,a new classification model is constructed,which has achieved good results in solving the problem of imbalanced multi classification.This paper uses a data set of medical diagnosis named Arrhythmia.The data set consists of 452 patient records,each with a total of 279 feature values,the aim is to classify the dataset into 16 categories.Among them,each category is unevenly distributed,more than half of the normal category accounted for the entire data set,some categories accounted for less than 1%.In the face of such a data set,a new classification model is established.The new classification model is based on the combination of radial basis function interpolation and logistic regression algorithm.In order to find the optimal classification model,four different radial basis functions,Gauss functions,Markoff distribution function,a polynomial function and the simplest functions,are used in the new classification model.Finally,it is found that the performance of the new classification model is the best and the classification accuracy is up to 76.01% when the penalty coefficient of is 1 and the coefficient of C is 10 in a polynomial function form.In order to facilitate the comparison,this paper also made a supplementary experiment,the radial basis function interpolation and logical regression applied to the data set,and the use of the previous reference to the data sets with different classifiers.Finally,it is found that the classification effect is the best.