Research on imbalanced data : based on SMOTE-AdaBoost algorithm

There are a lot of imbalanced data in many fields such as finance, information security and industrial system. How to extract valuable information from imbalanced data is a research hotspot and focus. In this paper, the imbalanced consumption data of credit card in UCI database are used. Oversamplin...

Full description

Saved in:
Bibliographic Details
Published in2019 3rd International Conference on Electronic Information Technology and Computer Engineering (EITCE) pp. 1165 - 1170
Main Authors Lv, Mengyu, Ren, Yi, Chen, Yufen
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.10.2019
Subjects
AdaBoost
Boosting
Classification algorithms
Data mining
Data processing
Error analysis
imbalanced data
SMOTE
SMOTE-AdaBoost
Training
Online AccessGet full text
DOI10.1109/EITCE47263.2019.9094859

Cover

More Information
Summary:There are a lot of imbalanced data in many fields such as finance, information security and industrial system. How to extract valuable information from imbalanced data is a research hotspot and focus. In this paper, the imbalanced consumption data of credit card in UCI database are used. Oversampling SMOTE method, AdaBoost algorithm and cost-sensitive algorithm are used to process the imbalanced data. Empirical results show that SMOTE-AdaBoost method is better than traditional AdaBoost method, and the cost-sensitive algorithm that increases the weight of the minority class samples is also higher than that of traditional AdaBoost method. Finally, this paper describes the challenges and future research directions of imbalanced data classification.
DOI:10.1109/EITCE47263.2019.9094859