A Python-based Data mining to Address Class Imbalance Problem

Orange canvas is an open source data mining tool that is based on Python scripting, visual programming and scientific computing. We developed analytical frameworks, which have advanced theoretical studies of practical learning methods, to address the class imbalance problem. In a two-class classification task, when the number of one class (majority) is greater than another (minority), this class is called imbalanced. The classification of this imbalanced class causes imbalanced distribution, poor predictive classification accuracy and a Class Imbalance Problem (CIP). We will focus on clarifying and writing a simple or clear Python script and visualize the frameworks of existing learning methods that address the CIP with well-known Synthetic over-sampling technique (SMOTE) based ensemble methods. The introduced orange workflows, Python scripting, and experimental results, will assist researchers and students to address the CIP simply. This study’s aim is to design innovative methods to address CIP.

کلیدواژه ها:

Python ، class imbalance problem ، SMOTE ، Orange Toolbox ، data mining