Volume 7 Issue 4 - November 2016

  • 1. An improved approach on class imbalance data using within-class minority oversampling technique

    Authors : Durga Parsad D, Dr K Nageswara Rao

    Pages : 156-164

    DOI : http://dx.doi.org/10.21172/1.74.020

    Keywords : Data Mining, Knowledge Discovery, Classification, Decision Tree, imbalance data, WIMOTE

    Abstract :

    Knowledge discovery from traditional or balance datasets can be done in an efficient way using the existing classification algorithms. The benchmark classification algorithms performance degrades when they are applied to the imbalance datasets. The reason is due to improper building of the predictive model using the imbalance datasets. In this paper, we propose a novel decision tree algorithm WithIn class Minority Oversampling TEchnique (WIMOTE) for efficient handling of imbalance data. The proposed WIMOTE approach uses the oversampling technique with unique statistical oversample strategy for removing misclassified and noisy instances in both majority and minority subset and oversamples the minority subset instances for data improvement. The experimental observation suggests that the proposed approach improves in terms of accuracy, AUC, Precision, Recall and F-measure with the benchmark SMOTE on 15 imbalance datasets from UCI repository.

    Citing this Journal Article :

    Durga Parsad D, Dr K Nageswara Rao, "An improved approach on class imbalance data using within-class minority oversampling technique ", https://www.ijltet.org/journal_details.php?id=909&j_id=3516, Volume 7 Issue 4 - November 2016, 156-164, #ijltetorg