Fast-CBUS: A fast clustering-based undersampling method for addressing the class imbalance problem

Ofek, Nir; Rokach, Lior; Stern, Roni; Shabtai, Asaf

doi:10.1016/j.neucom.2017.03.011

Cited by 119 publications

(61 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…On other hand Undersampling decrements the data count of wrong class so that right class is highlighted. RUS [16], CBUS [17], DSUS [18] and F-CBUS [19] are methods under it. Hybrid sampling is combination of oversampling and under sampling.…”

Section: Fig 4 Imbalance Datasetmentioning

confidence: 99%

A Recapitulation of Imbalanced Data

Layaq*,

Manjula

2020

IJITEE

View full text Add to dashboard Cite

In today’s authentic universe almost all applications are imbalanced. Data imbalance is growing faster than ever before as many systems are interested in extracting knowledge from lake of data. Imbalance issue occurs because required data is very rare and using that rare data if classification is done we may lead to inaccurate result. In few sensitive imbalance cases like medical and finance if classification is done health and wealth both will get a huge lost. It is observed that big data and imbalance issue are having hand in hand relationship. So, imbalance data is gaining much importance in data science. It is predicted that by the year 2020, about 1.7MB of lake of information will be created every second by each device due to development in science and technology. Almost this lake of information generated will be imbalanced. So, in this paper we will define big data and imbalanced data, how there are related to each other, some of the reasons why imbalance data problems are occurring, various areas where imbalance issues is been effecting, current four machine learning methods for imbalanced data (Data based method, Algorithm based method, Cost sensitive method and ensemble methods), overall performance evaluation of imbalance methods are done using a comparison chart and interpreting achievements of imbalanced data using confusion matrix, Combined evaluation measures (G-means, F-Measure, Balanced Accuracy, Youden Index and Matthews’s correlation coefficient) and Graphical performance evaluation using Receiver operating characteristic (ROC) curve and Area under the curve (AUC) and lastly, considering of result of various imbalance methods.

show abstract

Section: Fig 4 Imbalance Datasetmentioning

confidence: 99%

A Recapitulation of Imbalanced Data

Layaq*,

Manjula

2020

IJITEE

View full text Add to dashboard Cite

show abstract

“…Clustering is a kind of unsupervised learning, which is used to process data by many researchers [10][11][12][13][14]. Ahmad et al [10] proposed a clustering algorithm based on K-means paradigm, which is suitable for data with mixed numeric and categorical features.…”

Section: Related Workmentioning

confidence: 99%

“…Ahmad et al [10] proposed a clustering algorithm based on K-means paradigm, which is suitable for data with mixed numeric and categorical features. The combination of clustering and resampling [12][13][14][15] tends to produce better results. CBU (Clustering-based Undersampling) proposed by Lin et al [12] combines K-means and undersampling strategy, K-means only clusters the majority class and the number of clusters is same as that of the minority samples, and then CBU uses the nearest neighbor of each cluster center to represent the whole cluster and combines them with the minority samples to form a balanced training set.…”

Section: Related Workmentioning

confidence: 99%

“…Because of the number of data samples in minority class is less, EKR only clusters the majority class into k sub-clusters. We use K-means [10][11][12][13][14] as the clustering algorithm and combine k sub-clusters with the minority class into k subsets separately. Then, if the number of data samples in sub-cluster is larger than that in the minority class, undersampling the sub-clusters so that the number is same as that in minority class.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Novel Ensemble Framework Based on K-Means and Resampling for Imbalanced Data

et al. 2020

View full text Add to dashboard Cite

Imbalanced classification is one of the most important problems of machine learning and data mining, existing in many real datasets. In the past, many basic classifiers such as SVM, KNN, and so on have been used for imbalanced datasets in which the number of one sample is larger than that of another, but the classification effect is not ideal. Some data preprocessing methods have been proposed to reduce the imbalance ratio of data sets and combine with the basic classifiers to get better performance. In order to improve the whole classification accuracy, we propose a novel classifier ensemble framework based on K-means and resampling technique (EKR). First, we divide the data samples in the majority class into several sub-clusters using K-means, k-value is determined by Average Silhouette Coefficient, and then adjust the number of data samples of each sub-cluster to be the same as that of the minority classes through resampling technology, after that each adjusted sub-cluster and the minority class are combined into several balanced subsets, the base classifier is trained on each balanced subset separately, and finally integrated into a strong ensemble classifier. In this paper, the extensive experimental results on 16 imbalanced datasets demonstrate the effectiveness and feasibility of the proposed algorithm in terms of multiple evaluation criteria, and EKR can achieve better performance when compared with several classical imbalanced classification algorithms using different data preprocessing methods.

show abstract

“…Among undersampling techniques, k-means has been widely utilised in recent literature [2,[10][11][12]. By applying k-means, the majority class is divided into clusters before undersampling is performed resulting in a more balanced and diversified class distribution of the data.…”

Section: Introductionmentioning

confidence: 99%

Overlap-Based Undersampling for Improving Imbalanced Data Classification

Vuttipittayamongkol

Elyan

Petrovski

et al. 2018

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Classification of imbalanced data remains an important field in machine learning. Several methods have been proposed to address the class imbalance problem including data resampling, adaptive learning and cost adjusting algorithms. Data resampling methods are widely used due to their simplicity and flexibility. Most existing resampling techniques aim at rebalancing class distribution. However, class imbalance is not the only factor that impacts the performance of the learning algorithm. Class overlap has proved to have a higher impact on the classification of imbalanced datasets than the dominance of the negative class. In this paper, we propose a new undersampling method that eliminates negative instances from the overlapping region and hence improves the visibility of the minority instances. Testing and evaluating the proposed method using 36 public imbalanced datasets showed statistically significant improvements in classification performance.

show abstract

Fast-CBUS: A fast clustering-based undersampling method for addressing the class imbalance problem

Cited by 119 publications

References 23 publications

A Recapitulation of Imbalanced Data

A Recapitulation of Imbalanced Data

A Novel Ensemble Framework Based on K-Means and Resampling for Imbalanced Data

Overlap-Based Undersampling for Improving Imbalanced Data Classification

Contact Info

Product

Resources

About