Monte Carlo oversampling algorithm based on Bootstrap method

The problem of imbalanced data has been a hot problem in the field of data mining, and the problem of classification accuracy is low due to the tendency of imbalanced data classification to minority class. In this paper, we propose a new Monte Carlo sampling algorithm based on Bootstrap method. Firstly, the empirical distribution function of the minority class data is first obtained by the Bootstrap method, and then the distribution function is interpolated to obtain the probability density function. In addition, the weights of the new data are determined according to the probability density function of the minority class data. We use the Monte Carlo algorithm to oversample minority class data to make imbalanced data comes into balanced data. Finally, six sets of data are selected from UCI and KEEL databases for comparative study, and the decision tree classifier is selected for classification. Compared with the proposed algorithm in the past, the new algorithm is more feasible through experimental analysis.