Data Mining on Imbalanced Data Sets

Classification of Imbalanced Data by Combining the Complementary Neural Network and SMOTE Algorithm

“…In recent years, many research groups have found that an imbalanced data set could be one of the obstacles for many Machine Learning (ML) algorithms [1], [2], [3], [4]. In the learning process of the ML algorithms, if the ratio of minority classes and majority classes is significantly different, ML tends to be dominated by the majority classes and the features of the minority classes are recognize slightly.…”

“…In order to evaluate the classification performance of an imbalanced data set, the conventional classification accuracy cannot be used for this purpose because the minority class has minor impact on the accuracy when compared to the majority class [4]. Therefore, alternative measures have to be applied.…”

“…According to Gu et al [4], there are two main approaches to deal with imbalanced data sets: data-level approach and algorithm approach. While the data-level approach aims to re-balance the class distribution before a classifier is trained, the algorithm level approach aims to strengthen the existing classifier by adjusting algorithms to recognize the smaller classes.…”

“…Therefore, alternative measures have to be applied. The Geometric mean (Gmean) and the area under the Receiver Operating Characteristic (ROC) curve have been applied to evaluate the classification performance for imbalanced data set [4]. They are good indicators for the class imbalance problem because they attempt to maximize and balance the performance of ML between the minority class and the majority class.…”

“…SMOTE is applied because it can create new instances rather than replicate the existing instances. SMOTE is also the successful over-sampling technique applied commonly to the class imbalanced problem in the literature [1], [4].…”