A Survey of Predictive Modeling on Imbalanced Domains
Open accessThe DOI matches its Crossref record; 261 of the 266 references it cites were found in a registry or an index.
Abstract
Many real-world data-mining applications involve obtaining predictive models using datasets with strongly imbalanced distributions of the target variable. Frequently, the least-common values of this target variable are associated with events that are highly relevant for end users (e.g., fraud detection, unusual returns on stock markets, anticipation of catastrophes, etc.). Moreover, the events may have different costs and benefits, which, when associated with the rarity of some of them on the available training data, creates serious problems to predictive modeling techniques. This article presents a survey of existing techniques for handling these important applications of predictive analytics. Although most of the existing work addresses classification tasks (nominal target variables), we also describe methods designed to handle similar problems within regression tasks (numeric target variables). In this survey, we discuss the main challenges raised by imbalanced domains, propose a definition of the problem, describe the main approaches to these tasks, propose a taxonomy of the methods, summarize the conclusions of existing comparative studies as well as some theoretical analyses of some methods, and refer to some related problems within predictive modeling.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Here, techniques such as under-sampling and synthetic data have been found to enhance model performance [16].”
From A Survey of Data Quality Requirements That Matter in ML Development Pipelines (2023) · cites this paper as [16]
In the citing paper
“Earlier trends of opportunistic data collection and post hoc justifications of large datasets are gradually moving towards a requirement for more deliberative data collection methods [35,36], sampling techniques [16] and minimal data architectures [57] to deliver better performance without reducing model accuracy.”
From A Survey of Data Quality Requirements That Matter in ML Development Pipelines (2023) · cites this paper as [16]
Citations
Cites
SMOTE synthetic minority over-sampling technique
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002) · Journal of Artificial Intelligence Research · doi:10.1613/jair.953 · Cited by 32,865
UCI Machine Learning Repository
Asuncion, A. (2007) · Medical Entomology and Zoology · Cited by 25,411
Classification and Regression Trees.
Gordon, A. D., Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984) · Biometrics · doi:10.2307/2530946 · Cited by 23,522
An introduction to ROC analysis
Fawcett, T. (2006) · Pattern Recognition Letters · doi:10.1016/j.patrec.2005.10.010 · Cited by 21,879
He, H., & Garcia, E. A. (2009) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2008.239 · Cited by 10,359
A systematic analysis of performance measures for classification tasks
Sokolova, M., & Lapalme, G. (2009) · Information Processing & Management · doi:10.1016/j.ipm.2009.03.002 · Cited by 7,515
The use of the area under the ROC curve in the evaluation of machine learning algorithms
Bradley, A. P. (1997) · Pattern Recognition · doi:10.1016/s0031-3203(96)00142-2 · Cited by 7,217
Proceedings of the 25th international conference on Machine learning - ICML '08
(2008) · doi:10.1145/1390156 · Cited by 6,680
The relationship between Precision-Recall and ROC curves
Davis, J., & Goadrich, M. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143874 · Cited by 6,559
Estimating the Support of a High-Dimensional Distribution
Schölkopf, B., Platt, J. C., Shawe-Taylor, J., Smola, A. J., & Williamson, R. C. (2001) · Neural Computation · doi:10.1162/089976601750264965 · Cited by 6,168 · cited 1×
“Developments made in this context include one-class SVMs (e.g., Sch¨olkopf et al. [2001], Manevitz and Yousef [2002], Raskutti and Kowalczyk [2004], Zhuang and Dai [2006a, 2006b], and Lee and Cho [2006]) and the use of an autoencoder (or autoassociator) (e.g., Japkowicz et al. [1995] and Japkowicz…”
Method
Research works printed but not linked: 52
- Akbani Rehan, 2004, Machine Learning: ECML
- Alejo Roberto, 2013, Rosa Maria Valdovinos, and J. Horacio Pacheco-Sánchez
- Alejo Roberto, 2007, Ramón Alberto Mollineda, and José Salvador Sánchez.
- Eleuterio Roberto Alejo, 2011, Vicente García Jiménez, and Rosa María Valdovinos Rosas.
- Attenberg Josh, Imbalanced Learning: Foundations
- Bi Jinbo, Proc. of the 20th Int. Conf. on Machine Learning. 43--50
- Branco Paula, Re-sampling Approaches for Regression Tasks under Imbalanced Domains. Master's thesis. Dept. Computer Science
- Breiman Leo, 1984, Stone
- Chawla Nitesh V., 2003, Bowyer
- Chen Chao, 2004, Using random forest to learn imbalanced data
- Chyi Yu-Meei, 2003, Department of Information Management
- Cieslak David A., 2008, Chawla
- Doucette John, 2008, Heywood
- Drummond Chris, Workshop on Learning from Imbalanced Datasets II, 11
- Elkan Charles, 2001, IJCAI'01: Proc. of 17th Int. Joint Conf. of Artificial Intelligence, 1
- Ferri César, 2005, Proceedings of the Second Workshop on ROC Analysis in Machine Learning. 33--40
- Ferri César, Proceedings of the 28th International Conference on Machine Learning (ICML-11)
- Forman George, 2004, Knowledge Discovery in Databases: PKDD
- He Haibo, 2008, IEEE International Joint Conference on Neural Networks, 2008. IJCNN 2008. (IEEE World Congress on Computational Intelligence). IEEE, 1322--1328
- Holte Robert C., IJCAI, 89
- 10.5555/1681514.1682710 · doi:10.5555/1681514.1682710
- Japkowicz Nathalie, 2000, AAAI Workshop on Learning from Imbalanced Data Sets, 68
- Japkowicz Nathalie, 2003, Workshop on Learning from Imbalanced Data Sets II, 1723
- Japkowicz Natalie, Imbalanced Learning: Foundations
- Jeatrakul Piyasak, 2010, Kok Wai Wong, and Chun Che Fung
- Joshi Mahesh V., Proceedings IEEE International Conference on Data Mining, 2001. ICDM 2001. IEEE, 257--264
- Kubat Miroslav, 1997, Proc. of the 14th Int. Conf. on Machine Learning. Morgan Kaufmann, 179--186
- Liu Alexander, 2007, Martin
- Liu Wei, 2010, Chawla
- Maloof Marcus A., 2003, ICML-2003 Workshop on Learning from Imbalanced Data Sets II, 2
- Mani Inderjeet, 2003, Proceedings of Workshop on Learning from Imbalanced Datasets.
- 10.5555/1314498.1314514 · doi:10.5555/1314498.1314514
- Pearson Ronald, 2003, Proceedings of the ICML, 3
- Provost Foster J, 1998, ICML'98: Proc. of the 15th Int. Conf. on Machine Learning. Morgan Kaufmann Publishers, 445--453
- Raeder Troy, 2012, Chawla
- Rita, 2003, Ribeiro and Luís Torgo
- Rijsbergen Cornelis V., 1979, Information Retrieval. Dept. of Computer Science
- Sánchez-Crisostomo Juan Pablo, 2014, Rosa María Valdovinos, and J. Horacio Pacheco-Sánchez
- Sánchez-Monedero Javier, 2013, Pedro Antonio Gutiérrez, and Cesar Hervás-Martínez
- Songwattanasiri Panote, 2010, Proceedings of the Annual International Conference on Computer Science Education: Innovation and Technology
- Torgo Luís, 2003, Ribeiro
- Torgo Luís, 2007, Ribeiro
- Wagstaff Kiri L., 2013, Gilmore
- Wallace Byron C., 2011 IEEE 11th International Conference on Data Mining (ICDM). IEEE, 754--763
- Wang Heng, 2015, Concept drift detection for imbalanced stream data. arXiv Preprint arXiv:1504.01044 · arXiv
- Wang Xiaoguang, Advances in Artificial Intelligence
- Weiss Gary M., Data Mining
- Weiss Gary M., Imbalanced Learning: Foundations
- Cheng, Proceedings of the 7th Australasian Data Mining Conference-Volume 87
- Wu Gang, ICML 2003 Workshop on Learning from Imbalanced Data Sets II
- 10.5555/951949.952181 · doi:10.5555/951949.952181
- Zhang Dongmei, 2011, A novel improved SMOTE resampling algorithm based on fractal, J. Comput. Inform. Syst., 7, 2204
Cited by
Items 601–700 of 1,000
Automated imbalanced classification via layered learning
Cerqueira, V., Torgo, L., Branco, P., & Bellinger, C. (2022) · Machine Learning · doi:10.1007/s10994-022-06282-w · Cited by 11
“1 3 1 Introduction In supervised learning, we face an imbalanced problem when the distribution of classes is significantly biased towards a particular value and at the same time, the least frequent class is also the most relevant one (Branco et al., 2016).”
BackgroundSelf-citationFinding the Proverbial Needle: Improving Minority Class Identification Under Extreme Class Imbalance
Geisler, T., Ray, H. E., & Xie, Y. (2023) · Journal of Classification · doi:10.1007/s00357-023-09431-5 · Cited by 7
“For example, Branco et al. (2016) stated that as classes are skewed in numbers and in misclassification costs, the minority class observations should compensate in their assigned importance to make up the lack of numbers within the dataset.”
BackgroundAnnotation Cost Efficient Active Learning for Content Based Image Retrieval
Henkel, J., Hoxha, G., Sümbül, G., Möllenbrok, L., & Demir, B. (2023) · IEEE International Geoscience and Remote Sensing Symposium (IGARSS) · doi:10.1109/igarss52108.2023.10281811 · Cited by 6
“Since this imbalance is also reflected in the training set T , we stabilize the training by oversampling the minority class, as proposed in [11].”
MethodZhao, H., Tang, T., Lu, Y., Li, X., Sun, L., Chen, S., Ma, L., Luo, Y., Wang, K., & Zhao, M. (2023) · Ophthalmology and Therapy · doi:10.1007/s40123-023-00841-7 · Cited by 6
“In the evaluation of models for imbalanced data-sets, the use of ACC measures alone may lead to incorrect decisions, and the classification error rate of minority classes will be much higher than that of majority classes [35–38].”
Umirbekov, A., Essery, R., & Müller, D. (2024) · Geoscientific model development · doi:10.5194/gmd-17-911-2024 · Cited by 9
Montes, A. C., Ashok, P., & van Oort, E. (2024) · SPE/IADC Drilling Conference · doi:10.2118/217737-ms · Cited by 7
Exploring Resampling with Neighborhood Bias on Imbalanced Regression Problems
Branco, P., Torgo, L., & Ribeiro, R. P. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-65340-2_42 · Cited by 6
“More recently, it was shown that the problem of imbalanced domains also arises in several other tasks, namely: regression, data streams or multi-target prediction tasks [3, 8].”
BackgroundSelf-citationDiscovering Minority Sub-clusters and Local Difficulty Factors from Imbalanced Data
Lango, M., Brzeziński, D., Firlik, S., & Stefanowski, J. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-67786-6_23 · Cited by 6
“Improving classifiers learned from class-imbalanced data has been a topic of growing research in recent decades and several specialized algorithms have been introduced [2,6].”
BackgroundImplicit Interaction Through Machine Learning: Challenges in Design, Accountability, and Privacy
McMillan, D. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-70284-1_27 · Cited by 6
Resampling with neighbourhood bias on imbalanced domains
Branco, P., Torgo, L., & Ribeiro, R. P. (2018) · Expert Systems · doi:10.1111/exsy.12311 · Cited by 6
“The experiments were carried out using the following R packages: performanceEstimation (Torgo, 2014) for the experimental infrastructure and UBL (Branco et al., 2016) for the implementation of the random undersampling and SMOTE resampling strategies.”
BackgroundMethodInfluentialSelf-citationSelecting local ensembles for multi-class imbalanced data classification
Krawczyk, B., Cano, A., & Woźniak, M. (2018) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2018.8489572 · Cited by 6
“CS [23] Clustering and Selection base classifier: MLP clustering: k-means # clusters: [1-20] # classifiers: [5-100]”
BackgroundMethodInfluentialLeveraging Ensemble Pruning for Imbalanced Data Classification
Krawczyk, B., & Woźniak, M. (2018) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc.2018.00084 · Cited by 6
“Imbalanced data is one of the major contemporary challenges in machine learning, despite more than two decades of progress in this field [1], [2].”
BackgroundA New Loss Function for Traffic Classification Task on Dramatic Imbalanced Datasets
Xu, L., Zhou, X., Lin, X., Ren, Y., Qin, Y., & Liu, J. (2020) · International Conference on Communications · doi:10.1109/icc40277.2020.9148638 · Cited by 8
“Paula et al. propose some strategies to solve this problem in [6], such as data pre-processing and hybrid methods.”
BackgroundA Machine Learning guided Rewriting Approach for ASP Logic Programs
Mastria, E., Zangari, J., Perri, S., & Calimeri, F. (2020) · Electronic Proceedings in Theoretical Computer Science · doi:10.4204/eptcs.325.31 · Cited by 8
“Thus, the quality of the model has been assessed by means of the F1-score defined as F1 = 2 × (Precision × Recall)/(Precision+Recall) which provides us with more information about the effectiveness of the model on correctly predicting the instances belonging to minority classes [3].”
MethodWang, Z., Tsai, C., & Lin, W. (2021) · Data Technologies and Applications · doi:10.1108/dta-01-2021-0027 · Cited by 6
Kutty, S. K., & Menon, G. C. (2022) · International Journal of Imaging Systems and Technology · doi:10.1002/ima.22740 · Cited by 8
In silico prediction of mosquito repellents for clothing application
Devillers, J., Sartor, V., Doucet, J. P., Panaye, A. D., & Devillers, H. (2022) · SAR and QSAR in environmental research · doi:10.1080/1062936x.2022.2062871 · Cited by 6
“The closer the value is to 0, the more balanced both individual rates [46,48].”
BackgroundSecurity of Data Science and Data Science for Security
Tellenbach, B., Rennhard, M., & Schweizer, R. (2019) · Springer eBooks · doi:10.1007/978-3-030-11821-1_15 · Cited by 6
A novel Dual-Branch Generative Adversarial Network for electrocardiogram data generation
Zhou, F., & Huang, T. (2025) · Digital Signal Processing · doi:10.1016/j.dsp.2025.105149 · Cited by 5
Paixao, L. D. A. G., Rogers, W. P., & de Jesus, E. B. (2025) · Mining · doi:10.3390/mining5020024 · Cited by 6
“Hence, when dealing with imbalanced datasets, several different metrics were suggested in the literature to balance the criteria and evaluate the performance of a classifier [15,26,27].”
Fathnejat, H., & Nava, V. (2025) · Engineering With Computers · doi:10.1007/s00366-025-02148-6 · Cited by 7
“Balanced Accuracy (B-ACC) [44, 45], Average Class Specific Accuracy (ACSA) [44, 46], and Geometric Mean (GM) [47] are metrics used for evaluating classification performance.”
Kim, K.-N., Cho, D.-G., & Lee, M. J. (2025) · Buildings · doi:10.3390/buildings15234343 · Cited by 5
Su, Y.-K., & Tseng, C.-J. (2026) · Sensors · doi:10.3390/s26165026 · Cited by 5
“As a result, creating a unified framework that integrates engineered features, imbalance-aware learning, and deployment-oriented benchmarking remains a key research avenue for industrial PdM [42].”
Yin, J., Gao, C., Cheng, M., Liang, Q., Xue, P., Hao, S., & Zhao, Q. (2023) · Frontiers in Earth Science · doi:10.3389/feart.2022.1106799 · Cited by 5
“The learning goal of minimizing the expected difference makes the learning algorithm pay more attention to the characteristics of high-frequency distribution samples, resulting in lower prediction accuracy for data with a small number of samples (Branco et al., 2016; 2018).”
BackgroundMethodInfluentialAdiani, D., Colopietro, K., Wade, J. W., Migovich, M., Vogus, T. J., & Sarkar, N. (2023) · Signals · doi:10.3390/signals4020019 · Cited by 9
“Instead of random oversampling (creating copies of the minority samples), which is known to lead to overfitting (model trains well on training data, but performs poorly on unseen or test data [64]) [65], we added some new data to the ft, fp, fa, and fe classes.”
MethodVisActive: Visual-concept-based Active Learning for Image Classification under Class Imbalance
Khaleel, M., Idris, A., Tavanapong, W., Pratt, J., Oh, J., & de Groen, P. C. (2023) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3617999 · Cited by 7
“Class imbalance is present in many application domains [16-19].”
BackgroundThe genetic framework of primary ciliary dyskinesia assessed by soft computing analysis
Pifferi, M., Boner, A. L., Cangiotti, A. M., Cudazzo, A., Maj, D., Gracci, S., Michelucci, A., Bertini, V., Piazza, M., Valetto, A., et al. (2024) · Pediatric Pulmonology · doi:10.1002/ppul.26842 · Cited by 6
“Furthermore, since the accuracy is not sufficient to determine the model performance in case of unbalanced datasets, 26 to visualize the behavior of the classifier we used two methods: the macroaveraging and microaveraging receiver operator characteristic (ROC) curve, and the confusion matrix…”
MethodA detailed study of resampling algorithms for cyberattack classification in engineering applications
Gutiérrez, Ó. M., Núñez, J. C. S., Ávila, M., & Lindo, A. C. (2024) · PeerJ Computer Science · doi:10.7717/peerj-cs.1975 · Cited by 6
Exploring the Impact of the NULL Class on In-the-Wild Human Activity Recognition
Cherian, J., Ray, S., Taele, P., Koh, J. I., & Hammond, T. (2024) · Sensors · doi:10.3390/s24123898 · Cited by 6
“Techniques designed to handle data imbalances in machine learning applications are applied as a preprocessing step (data level), as part of the training of the model (algorithm level), or both (hybrid) [47–49].”
MethodSparse feature selection and rare value prediction in imbalanced regression
Guan, Y., & Fu, G. (2024) · Information Sciences · doi:10.1016/j.ins.2024.121145 · Cited by 8
Moawed, S. A., Mahrous, E., Elaswad, A., Gouda, H. F., & Fathy, A. (2024) · BMC Veterinary Research · doi:10.1186/s12917-024-04234-1 · Cited by 7
“We opted against Synthetic Minority Over-sampling Technique (SMOTE) due to its potential drawbacks for sparse minority classes [39] and using SMOTE can disrupt the class separation that discriminant analysis relies on [40].”
MethodGuo, Y., Kou, Y., Yi, L., & Fu, G. (2025) · Journal of Chemometrics · doi:10.1002/cem.70029 · Cited by 6
“…class samples makes it difficult for the model to learn the characteristics of minority class samples, thereby reducing the model's generalization [20], (2) the dataset contains a lot of noisy data, and the minority class samples have weak resistance to noise, resulting in poor differentiation…”
Hu, Y., Tsai, C., & Wang, P.-T. (2025) · Annals of Operations Research · doi:10.1007/s10479-025-06706-5 · Cited by 5
“Three primary types of data re-sampling techniques have been widely studied: under-sampling, over-sampling, and hybrid sampling (Branco et al., 2016; Galar et al., 2012; He et al., 2009; Leevy et al., 2018).”
wCM based hybrid pre-processing algorithm for class imbalanced dataset
Singh, N. P., Saha, A., & Gosain, A. (2021) · Journal of Intelligent & Fuzzy Systems · doi:10.3233/jifs-210624 · Cited by 5
“Hence, the probability of misclassification of the minority class is high compared to the majority class [1].”
Assembly Quality Detection Based on Class-Imbalanced Semi-Supervised Learning
Lu, Z., Jiang, J., Cao, P., & Yang, Y. (2021) · Applied Sciences · doi:10.3390/app112110373 · Cited by 5
“We used two common metrics in class-imbalanced learning to evaluate the performance of the model on the test set: balanced accuracy (bACC) [37] and geometric mean score (GM) [38], which are arithmetic and geometric mean scores, respectively, defined as”
MethodRobust and efficient COVID-19 detection techniques: A machine learning approach
Hasan, M. M., Murtaz, S. B., Islam, M. U., Sadeq, M. J., & Uddin, J. (2022) · PLoS ONE · doi:10.1371/journal.pone.0274538 · Cited by 7
Kurasawa, H., Waki, K., Chiba, A., Seki, T., Hayashi, K., Fujino, A., Haga, T., Noguchi, T., & Ohe, K. (2022) · JMIR Bioinformatics and Biotechnology · doi:10.2196/37951 · Cited by 6
“According to survey papers [41-43] on biased data, sampling has often been attempted as a way to solve this problem [44,45].”
BackgroundWankmüller, S. (2022) · Journal of Computational Social Science · doi:10.1007/s42001-022-00191-7 · Cited by 6
“The statistical learning community has devised a large spectrum of approaches to deal with imbalanced classification problems (for an overview see [24]).”
BackgroundReal-Time Privacy-Preserving Fall Detection using Dynamic Vision Sensors
Prasad, S. S., Mehta, N. K., Banerjee, A., Kumar, H., Saurav, S., & Singh, S. K. (2022) · IEEE India Conference · doi:10.1109/indicon56171.2022.10039894 · Cited by 7
“Measuring the performance of a model in terms of accuracy when the dataset is having class imbalance leads to the accuracy paradox [22].”
BackgroundExtreme weather impacts do not improve conflict predictions in Africa
Michelini, S., Šedová, B., Schewe, J., & Frieler, K. (2023) · Humanities and Social Sciences Communications · doi:10.1057/s41599-023-01996-1 · Cited by 7
Zhang, P., Liu, X., & Pu, K. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3339682 · Cited by 6
“Also, G-mean, which balances the ability of the model to correctly classify the two categories, is used to assess the classification results for data imbalanced problems [47], [48].”
MethodA hybrid model: PNM for improving prediction capability of classifier
Mehrotra, S., Muttum, V. K., Krishna, R. V., Kumar, V., & Varish, N. (2023) · International Journal of Information Technology · doi:10.1007/s41870-023-01609-9 · Cited by 5
“[4] presented a survey of predictive models for class imbalance data [11].”
Xie, B., Chen, B., Ma, J., Chen, J., Zhou, Y., Han, X., Xiong, Z., Yu, Z., & Huang, F. (2024) · Food Analytical Methods · doi:10.1007/s12161-024-02574-0 · Cited by 5
Estévez, V., Mattbäck, S., Boman, A., Liwata-Kenttälä, P., Björk, K.-M., & Österholm, P. (2024) · Geoderma · doi:10.1016/j.geoderma.2024.116916 · Cited by 9
FHRDiff: Leveraging Diffusion Models for Conditional Fetal Heart Rate Signal Generation
Li, X., Lu, Y., Ngiam, K. Y., Yu, Z., & Furqan, M. S. (2024) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm62325.2024.10821853 · Cited by 7
RACH Traffic Prediction in Massive Machine Type Communications
Mehri, H., Mehrpouyan, H., & Chen, H. (2025) · IEEE Transactions on Machine Learning in Communications and Networking · doi:10.1109/tmlcn.2025.3542760 · Cited by 8
“The second metric is recall and is calculated as [56]: where, ‘ False negatives ’ are bursty samples that incorrectly detected as non-bursty samples.”
InfluentialData Augmentation with Variational Autoencoder for Imbalanced Dataset
Stocksieker, S., Pommeret, D., & Arthur Charpentier, A. (2025) · Lecture notes in computer science · doi:10.1007/978-981-96-6579-2_24 · Cited by 6
Adapted pruning scheme for the framework of imbalanced data-sets
Chaabane, I., Guermazi, R., & Hammami, M. (2017) · Procedia Computer Science · doi:10.1016/j.procs.2017.08.060 · Cited by 5
A Hybrid Sampling Method Based on Safe Screening for Imbalanced Datasets with Sparse Structure
Shi, H., Gao, Q., Ji, S., & Liu, Y. (2018) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2018.8489569 · Cited by 5
“This may eliminate some useful instances for classification, leading to a worse performance [8].”
BackgroundExploiting label information to improve auto-encoding based classifiers
Sánchez-Morales, A., Sancho‐Gómez, J., & Figueiras-Vidal, A. R. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.08.055 · Cited by 5
Asymmetric label switching resists binary imbalance
Gutiérrez-López, A., González-Serrano, F. J., & Figueiras-Vidal, A. R. (2020) · Information Fusion · doi:10.1016/j.inffus.2020.02.004 · Cited by 5
AI Tool with Active Learning for Detection of Rural Roadside Safety Features
Yi, H., Bizon, C., Borland, D., Watson, M., Satusky, M., Rittmuller, R., Radwan, R. A., Srinivasan, R., & Krishnamurthy, A. K. (2021) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata52589.2021.9671360 · Cited by 7
“Branco et al. [32] presented a survey of unbalanced learning that discussed the main challenges raised by imbalanced data distributions and approaches to address these challenges.”
BackgroundA Hybrid Feature Ranking Algorithm for Assisted Reproductive Technology Outcome Prediction
Kothandaraman, R., Suruliandi, A., & Raja, S. P. (2022) · Brazilian Archives of Biology and Technology · doi:10.1590/1678-4324-2022210605 · Cited by 5
Online Anomaly Explanation: A Case Study on Predictive Maintenance
Ribeiro, R. P., Mastelini, S. M., Davari, N., Aminian, E., Veloso, B., & da Gama, J. M. P. (2023) · Communications in computer and information science · doi:10.1007/978-3-031-23633-4_25 · Cited by 5
Self-citationAzeroual, S., Ben-Bouazza, F.-E., Naqi, A., & Sebihi, R. (2023) · Lecture notes in networks and systems · doi:10.1007/978-3-031-35248-5_39 · Cited by 5
Fathi, K., Sadurski, M., Kleinert, T., & van de Venn, H. W. (2023) · International Conference on Control, Automation and Systems (ICCAS) · doi:10.23919/iccas59377.2023.10316874 · Cited by 5
Gerli, C., Rosier, S. H. R., Gudmundsson, G. H., & Sun, S. (2024) · The Cryosphere · doi:10.5194/tc-18-2677-2024 · Cited by 5
Bolívar, A., García, V., Alejo, R., Florencia-Juárez, R., & Sánchez, J. S. (2024) · Applied Sciences · doi:10.3390/app14135845 · Cited by 6
Proposing Logical Table Constructs for Enhanced Machine Learning Process
Uddin, M. F., Rizvi, S. S., & Razaque, A. (2018) · IEEE Access · doi:10.1109/access.2018.2866046 · Cited by 5
Neural Networks for Predicting the Output of wind flow Simulations Over Complex Topographies
Mayo, M., Wakes, S., & Anderson, C. S. (2018) · IEEE International Conference on Big Knowledge (ICBK) · doi:10.1109/icbk.2018.00032 · Cited by 5
“Imbalanced data is well known to cause problems when training and evaluating machine learning models and therefore it must be handled carefully [32].”
BackgroundMethodImbalanced Data Classification Based on Hybrid Methods
Zhang, N.-N., Shao-zhen, Y., & Chien, T. (2018) · International Conference Big Data Research · doi:10.1145/3291801.3291812 · Cited by 5
“Random under-sampling can eliminate important majority class instances and randomly duplicating minority class instances does not diversify the information that facilitates classification; thus, overfitting can occur [10].”
BackgroundNg, W. W. Y., Zhang, Y., Zhang, J., Wang, D. D., & Wang, F. L. (2019) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2019.2922340 · Cited by 5
“Class imbalance occur when one class is severely un-derrepresented [12], that is, one class contains much less samples than the other class.”
BackgroundBiased Resampling Strategies for Imbalanced Spatio-Temporal Forecasting
Oliveira, M., Moniz, N., Torgo, L., & Costa, V. S. (2019) · IEEE International Conference on Data Science and Advanced Analytics (DSAA) · doi:10.1109/dsaa.2019.00024 · Cited by 5
Self-citationFiltering-Based Instance Selection Method for Overlapping Problem in Imbalanced Datasets
Rubbo, M., & da Silva, L. A. (2021) · J — Multidisciplinary Scientific Journal · doi:10.3390/j4030024 · Cited by 5
“From among the techniques available in the relevant literature, it is possible to highlight the random methods, distance methods, and clustering and evolutionary algorithms [15].”
BackgroundIn-Between Layers Modular Residual Neural Network for the Classification of Images
Valova, I., Harris, C., Gueorguieva, N., & Mai, T. (2021) · Procedia Computer Science · doi:10.1016/j.procs.2021.05.023 · Cited by 5
Addressing Fairness, Bias and Class Imbalance in Machine Learning: the FBI-lossPreprint
Ferrari, E., & Bacciu, D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.06345 · Cited by 12
“We stress that, despite the adoption of Accuracy metric in CI problems is deprecated because its value is biased by the disproportion of the class frequencies (Branco et al., 2016), we avoid this effect by computing the Accuracy separately for the UnderG and OverG groups.”
MethodWang, K., Shah, S. B., Chen, H., Perrault, A., Doshi‐Velez, F., & Tambe, M. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.03279 · Cited by 5
“…loss on the training MDPs does not always imply a high off-policy evaluation on the training MDPs in Figure 2(b) due to the misalignment of predictive accuracy and decision quality, which is consistent with other studies in mismatch of predictive loss and evaluation metric [6, 14, 15, 20].”
BackgroundResultDrug target prediction through deep learning functional representation of gene signaturesPreprint
Zhou, Y., Chen, H., King, F. J., Zhou, B., Wang, Y., Canedy, C. J., Hayashi, J., Zhong, Y., Chang, M. W., Pache, L., et al. (2023) · Research Square · doi:10.21203/rs.3.rs-3371688/v1 · Cited by 5
The ROC-AUC Accurately Assesses Imbalanced DatasetsPreprint
Richardson, E., Trevizani, R., Greenbaum, J., Carter, H., Nielsen, M., & Peters, B. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4655233 · Cited by 5
Exploring Resampling Techniques in Credit Card Default PredictionPreprint
Lokanan, M. E. (2024) · Research Square · doi:10.21203/rs.3.rs-4087259/v1 · Cited by 9
At-Home Breath Data Collection for Signatures of Type 2 Diabetes: A Pilot Clinical Study
Naser, S., Roberts, D. A., & Shrestha, S. (2025) · Biosensors · doi:10.3390/bios15030149 · Cited by 5
“We employed two key metrics: the F1 score and class balance ratio [19].”
Phung, K., Aydın, M. E., & Ogunshile, E. K. A. (2025) · Concurrency and Computation Practice and Experience · doi:10.1002/cpe.70472 · Cited by 4
The Use of Machine Learning in Aluminum and Magnesium Matrix Composites: A Review
Aydın, F., & Duru, A. (2026) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-026-10511-z · Cited by 7
Leitner, M.-L., & Arendasy, M. E. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1746479 · Cited by 7
“Without corrective measures, such as resampling techniques, class weighting, or cost-sensitive learning, predictive performance for the minority class may be substantially compromised (Chen et al., 2004; Branco et al., 2016).”
Zhang, J., Wang, X., Wang, D., Zheng, Z., Wang, H., & La, M. (2026) · mSystems · doi:10.1128/msystems.00800-25 · Cited by 6
“Early successes have demonstrated that reliable detection of members of the “rare biosphere” is achievable without the need for arbitrary abundance thresholds (86). and 3), resulting in ecologically significant rare taxa being overlooked (129, 130).”
Perturbation-based oversampling technique for imbalanced classification problems
Zhang, J., Wang, T., Ng, W. W. Y., & Witold Pedrycz, C. (2022) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-022-01662-z · Cited by 4
“It is widely believed that traditional classifiers directly trained from imbalanced data are prone to misclassify minority instances [7].”
BackgroundVijayvargiya, A., Sinha, A., Gehlot, N., Jena, A., K., R., & Moran, K. (2024) · PLoS ONE · doi:10.1371/journal.pone.0301263 · Cited by 7
A framework to undersample and refine the synthetic minority set
Sadhukhan, P. (2025) · Applied Soft Computing · doi:10.1016/j.asoc.2025.113095 · Cited by 5
Enhancing Human Activity Recognition With Iterative Adaptive Approach in Assisted Living
Akaydin, A., Li, Z., Robertson, A., Romain, O., & Kernec, J. L. (2025) · IEEE Sensors Journal · doi:10.1109/jsen.2025.3597939 · Cited by 6
“Restrictions apply. complementary resampling strategies: the synthetic minority over-sampling technique (SMOTE) [35] and a hybrid random over-sampling and random under-sampling approach (ROS-RUS) [36].”
AUTOMATIC CHANGE DETECTION OF DIGITAL MAPS USING AERIAL IMAGES AND POINT CLOUDS
Dahle, F., Ohori, K. A., Agugiaro, G., & Briels, S. (2021) · The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciences · doi:10.5194/isprs-archives-xliii-b2-2021-457-2021 · Cited by 4
Artifact Detection in Invasive Blood Pressure Data using Forecasting Methods and Machine Learning
Wu, M., Branco, P., Ke, J. X. C., & Macdonald, D. B. (2020) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm49941.2020.9313540 · Cited by 4
“These metrics are recommended as suitable for performance evaluation in the context of imbalanced domains [18] which is the case of our data set (artifacts are under-represented in the available data and are the most important class to detect).”
BackgroundSelf-citationImproving User Environment Detection Using Context-Aware Multi-Task Deep Learning in Mobile Networks
Morel, M.-L. A., Saffar, I., Singh, K., Hamidèche, S. A., & Viho, C. (2022) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2022.3205696 · Cited by 7
“Actually, different techniques can be found in literature to solve [36].”
Mecati, M., Torchiano, M., Vetrò, A., & Martin, J. C. D. (2023) · IEEE Access · doi:10.1109/access.2023.3252370 · Cited by 10
“diagnostics, and a number of other fields [24]: often negatively labeled instances significantly outnumber positively labeled instances, but the latter are associated with the most relevant events for end users (e.”
BackgroundMethodLi, R., Millist, L., Foster, E., Yuan, X., Guvenc, U., Radfar, M., Marendy, P., Ni, W., O'Brien, T. J., & Casillas-Espinosa, P. M. (2023) · Epilepsy Research · doi:10.1016/j.eplepsyres.2023.107181 · Cited by 4
Sarigiannis, D., Atzemi, M., Oke, J. B., Christofa, E., & Gerasimidis, S. (2024) · Transportation Research Record Journal of the Transportation Research Board · doi:10.1177/03611981231217497 · Cited by 5
Finding score-based representative samples for cancer risk prediction
Liao, J., Luo, H., Yan, X., Ye, T., Huang, S., & Liu, L. (2024) · Pattern Recognition · doi:10.1016/j.patcog.2024.110767 · Cited by 4
An ensemble model for addressing class imbalance and class overlap in software defect prediction
Dar, A. W., & Farooq, S. U. (2024) · International Journal of Systems Assurance Engineering and Management · doi:10.1007/s13198-024-02538-x · Cited by 4
“Data-level techniques are the more often used techniques as these techniques are less complex and are easy to implement on any dataset (Branco et al. 2016).”
A similarity-based semi-supervised algorithm for labeling unlabeled text data
Potshangbam, K. S., & Singh, K. N. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.128941 · Cited by 5
Cui, X., Coenen, F., & Bollegala, D. (2017) · e-space (Manchester Metropolitan University) · Cited by 4
Learning Through Utility Optimization in Regression Tasks
Branco, P., Torgo, L., Ribeiro, R. P., Frank, E., Pfahringer, B., & Rau, M. M. (2017) · IEEE International Conference on Data Science and Advanced Analytics (DSAA) · doi:10.1109/dsaa.2017.63 · Cited by 4
“, [18], [19]).”
Self-citationMinority Oversampling Using Sensitivity
Zhang, J., Wang, T., Ng, W. W. Y., Pedrycz, W., Zhang, S., & Nugent, C. (2020) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn48605.2020.9206601 · Cited by 4
“Class imbalance occurs when a class consists of much less examples than the other class [1] and it is prevalent in many real-world applications, for examples anomaly detection [2], electricity pricing [3], and diagnosis of rare diseases [4].”
BackgroundJamaluddin, A. H., & Mahat, N. I. (2020) · Journal of Information and Communication Technology · doi:10.32890/jict.20.1.2021.6358 · Cited by 4
Learning algorithms to deal with failures in production planning
Alves, F. F., Nogueira, T. H., & Ravetti, M. G. (2022) · Computers & Industrial Engineering · doi:10.1016/j.cie.2022.108231 · Cited by 4
Li, X., Zhang, A., Al-Zaidy, R. A., Rao, A., Baral, S., Bao, L., & Giles, C. L. (2022) · PLoS ONE · doi:10.1371/journal.pone.0270034 · Cited by 6
“As our study contains two classes of data that are imbalanced, AUC-ROC is not capable of fully reflecting model performance [29].”
Devillers, J., Sartor, V., & Devillers, H. (2022) · SAR and QSAR in environmental research · doi:10.1080/1062936x.2022.2124014 · Cited by 6
“The closer the value is to 0, the more balanced both individual rates [48,49].”
BackgroundDiscriminative Singular Spectrum Classifier with applications on bioacoustic signal recognition
Gatto, B. B., Colonna, J. G., Santos, E. M. D., Koerich, A. L., & Fukui, K. (2022) · Digital Signal Processing · doi:10.1016/j.dsp.2022.103858 · Cited by 5
Automatic identification of conodont species using fine-grained convolutional neural networks
Duan, X. (2023) · Frontiers in Earth Science · doi:10.3389/feart.2022.1046327 · Cited by 6
“Therefore, in the case of unbalanced datasets, two metrics, ROC and AUC, are usually used instead of (Branco et al., 2016).”
BackgroundChang, L., Kulshrestha, A., Zhang, B., & Zhang, X. (2023) · Remote Sensing · doi:10.3390/rs15061571 · Cited by 4
Golukhova, E. Z., Keren, M. A., Zavalikhina, T. V., Bulaeva, N. I., Акатов, Д. С., Sigaev, I. Y., Яхяева, К. Б., & Kolesnikov, D. A. (2023) · Russian Journal of Cardiology · doi:10.15829/1560-4071-2023-5211 · Cited by 4
A Submodular Optimization Framework for Imbalanced Text Classification With Data Augmentation
Alemayehu, E., & Fang, Y. (2023) · IEEE Access · doi:10.1109/access.2023.3267669 · Cited by 4
Metadata
- Authors
- Branco, P., Torgo, L., & Ribeiro, R. P.
- Affiliations
- Universidade do Porto (PT)
- LIAAD-INESC TEC, DCC-Faculty of Sciences, University of Porto, Porto, Portugal
- Venue
- ACM Computing Surveys
- Published
- 2016-08-13
- Volume
- 49
- Issue
- 2
- Pages
- 1-50
- ISSN
- 0360-0300 (ISSN-L)
- 1557-7341
- OpenAlex
- W2490420619
- Semantic Scholar
- 736018160f8e6f6bd9b2a9101cd13ee67e573097
- Cited by
- 1,129
- References
- 266
- Citations by year
- 2026: 101 · 2025: 113 · 2024: 167 · 2023: 167 · 2022: 176 · 2021: 137 · 2020: 96 · 2019: 72 · 2018: 60 · 2017: 37 · 2016: 3
- FWCI
- 47.12
- Type
- journal article
- Language
- en
- Open access
- Locations
- ACM Computing Surveys (publishedVersion)
- Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) (submittedVersion, other-oa)
- Funders
- North Portugal Regional Operational Programme (Project NORTE-07-0124-FEDER-000059)
- European Regional Development Fund (POCI-01-0145-FEDER-006961)
- Fundação para a Ciência e a Tecnologia (PD/BD/105788/2014, SFRH/BSAB/113896/2015)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Imbalanced Data Classification Techniques
- Social Sciences › Business, Management and Accounting › Marketing › Customer churn and segmentation
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Keywords
- class imbalance, predictive modeling, rare event prediction, cost-sensitive learning, classification, regression analysis, fraud detection
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.