Adaptive Semi-Unsupervised Weighted Oversampling with Sparsity Factor for Imbalanced Biomedical Data

Ali, Haseeb; Samat, Nurul Ashikin; Ashgher, Hafiz Maaz

doi:10.30880/jscdm.2020.01.01.003

Cited by 1 publication

(1 citation statement)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Data imbalance results in misclassification or modeling errors where data is irrelevant and results in poor classification modeling [4]. The imbalance of data greatly affects several classifications such as credit data [5], stroke data , [6]online [7]news data , biomedical data [8], diarrhea case data for toddlers [9], poor household classification data [10], and other data.…”

Section: Introductionmentioning

confidence: 99%

Evaluate of Random Undersampling Method and Majority Weighted Minority Oversampling Technique in Resolve Imabalanced Dataset

Untoro,

Yusuf

2023

ITJRD

View full text Add to dashboard Cite

Classification is a model for making predictions based on existing data. Imbalanced data leads to misclassification or modeling errors where the data is not relevant and results in poor classification modeling. A poor classification model is caused by imbalanced data in the classification label, and there is a need for data balancing as a solution to resolve this issue. The methods used to handle data imbalance are Random Undersampling and MWMOTE. The goal is to see the implementation of Random Undersampling and MWMOTE working well in addressing the imbalanced dataset and to know the performance and accuracy in modeling. The dataset used is an open source dataset from Kaggle consisting of Diabetes data, Bank Turnover data, Stroke data, and Credit Card data with various data ratios, with the goal of addressing the problem of imbalanced data. Model evaluation was performed using the confusion matrix and decision tree algorithm by looking at the precision, recall, f-measure, and accuracy values from the Random Undersampling and MWMOTE methods. Random Undersampling can address the problem of imbalanced data with a precision of 76.28%, recall of 76.74%, f-measure of 76.48%, and accuracy of 76.21%. MWMOTE can address the problem of imbalanced data with a precision of 86.04%, recall of 87.30%, f-measure of 86.66%, and accuracy of 86.61%. It can be concluded that the MWMOTE method is better than the Random Undersampling method because the average evaluation of the confusion matrix of the Random Undersampling method is smaller than the MWMOTE method.

show abstract