A Review on Handling Multiclass Imbalanced Data Classification In Education Domain
Rose Mary Mathew, Ramalingam Gunasundari · 2021 International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE) · 2021
Multiclass imbalanced data classification is off late considered as one of the major pain points in the machine learning domain. This has changed from a need to necessity as the requirement to process the real-life skewed data has increased exponentially. The problem arises when you have a big parity in the number of examples for a single class, when compared to others. The challenges with most of the existing algorithms is that they will only rightly identify the class with the greater number and fail to identify the minority classes. Educational Data Mining is an emerging field for studying data in the educational context by applying various algorithms in machine learning. In educational sector most of the data is skewed and only a few works are there for handling it as balanced. In this study, we address multiclass imbalanced data first and identify the various methods that deal with this and a survey about the usage of these methods in the educational data.