Investigating the effect of correlation based feature selection on breast cancer diagnosis using artificial neural network and support vector machines

The breast cancer is one of the most popular cause of death among women. It is also one of the diseases that can be cured and has high healing chances when it is detected in the early stages [1]. Detecting the cancer and differentiating between the diagnosis that affirm whether a patient has breast cancer or not has been considered as a big challenge. In order to have an accurate diagnosis, Support Vector Machine (SVM) and Artificial Neural Network (ANN) have been selected in many research papers to solve this problem with high classification accuracy. In this paper the breast cancer diagnosis is addressed using SVM and ANN combined with feature selection. The feature selection is based on the correlation coefficient of each feature against the target class where different feature subsets are used. The model is tested on the popular Wisconsin Diagnosis Breast Cancer (WDBC) dataset to conduct the experiments. 10- Fold Cross validation has been used for data partitioning while developing the model and the outcome indicates better classification accuracy. As for comparison between SVM and ANN, empirical studies outcome indicated that SVM outperformed ANN with classification accuracy of 97.14 and 96.71 respectively.