Performance of synthetic minority oversampling technique on imbalanced breast cancer data

Classification, one of the important data mining techniques plays an important role in classifying the data for knowledge discovery. The data sets contain some redundant and unnecessary attributes which mislead the classifiers. Feature Extraction techniques such as Principal Component Analysis (PCA) can be used to overcome high-dimensionality problem. Class imbalance is another problem which leads difficulty in learning for classifiers. Hence imbalanced datasets should be restructured using the resampling techniques. Synthetic Minority Oversampling Technique (SMOTE) can control the number of examples and distribution to achieve the purpose of balancing the dataset through synthetic new examples. In this study the performance of SMOTE with five classifiers on breast cancer data sets are analyzed.