A Review on Data Balancing Techniques and Machine Learning Methods

• Published in: 2023 5th International Conference on Smart Systems and Inventive Technology (ICSSIT) pp. 1004 - 1008
• Main Authors: Parmar, Gaurav, Gupta, Rimi, Bhatt, Tejas, Sahani, G.J., Panchal, Brijeshkumar Y., Patel, Hiren
• Format: Conference Proceeding
• Language: English
• Published: IEEE 23-01-2023
• Subjects: Class Bias; Data mining; Distributed databases; Machine learning; Over Sampling; Parallelization; Protected Attributes; Under Sampling
• ISSN: 2832-3017
• DOI: 10.1109/ICSSIT55814.2023.10061154

Researchers in the discipline of Data Mining (DM) occasionally disregard the need of ensuring a dataset is evenly distributed. As such, it may have a major impact on how things are ultimately sorted. Most classifiers function on the concept that the data is roughly normally distributed. Therefore, the categorization approach is no longer as successful as it once was and fixing this is essential. This research aims to construct an evaluation of several methods for producing labels for minority classes to prevent the model from developing bias toward protected attributes. In this study, different researcher's work learned to determine the most effective method for disseminating the data. There will be a discussion of ways to enhance categorization abilities in the conclusion. The future of data-balancing research is determined by weighing the relative merits of the several possible methodological amalgamations.