|Communications on Applied Electronics
|Foundation of Computer Science (FCS), NY, USA
|Volume 7 - Number 36
|Year of Publication: 2021
|Authors: Varsha Babar
Varsha Babar . Classification of Imbalanced Data of Medical Diagnosis using Sampling Techniques. Communications on Applied Electronics. 7, 36 ( May 2021), 7-12. DOI=10.5120/cae2021652883
When there is gigantic difference between the ratio of two classes in the classification algorithms, then the classifier may tend to favor the instances of majority class whereas, it becomes difficult for the classifier to learn the minority class samples. Either, undersampling is used or oversampling is used for this imbalance but, most of the undersampling techniques does not consider distribution of information among the classes while the oversampling technique leads overfitting of the trained model. So, to resolve this issue integration of undersampling as well as oversampling technique can be done. Majority class samples can be undersampled using a new approach, namely, MLP-based undersampling technique (MLPUS). Majority Weighted Minority Oversampling Technique (MWMOTE) can be used for generating the synthetic samples for minority class. The main objective is to handle the imbalance classification problem occurring in the medical diagnosis of rare diseases and combines the benefits of both undersampling and oversampling Experiments are performed on 7 real world data sets for the evaluation of proposed framework’s performance.