APPARATUS FOR RESOLVING CLASS IMBALANCE PROBLEM USING FCM AND SMOTE AND METHOD THEREOF

摘要

The present invention relates to an apparatus for resolving a class imbalance problem using fuzzy C-means clustering (FCM) and a synthetic minority oversampling technique (SMOTE) and a method thereof which can prevent biased learning by combining the FCM and the SMOTE. The apparatus comprises: a classification module to divide learning data for implementing regression analysis and classification techniques into majority class data and minority class data when the learning data are inputted as source data; a clustering module to apply FCM to the minority class data divided by the classification module; a parameter setting module to use extended gap statistics to derive an optimal number of clusters; and a data generation module to apply a SMOTE to generate synthesis data.;COPYRIGHT KIPO 2019