A Novel SVM-KNN Classifier for Cervical Cancer Diagnosis using Feature Reduction and Imbalanced Learning Techniques

Cervical cancer is one sort of prenatal tumors and a large portion of the complexities of cancer threatening causes to deaths which were identified in most of the countries. There are different risk factors related to cancer threatening development. The number of methodologies developed to predict this cancer such as Decision Tree (DT), K-nearest neighbors (KNN), Support vector machine (SVM), Random Forest (RF), Logistic Regression (LR), Principal Component Analysis (PCA) and Logistic Regression (LR). However, it is observed that most of the medical data suffer from class imbalance issues. The work in this paper proposed an ensemble classifier using SVM and KNN with an oversampling technique called Synthetic Minority Oversampling Technique (SMOTE) for Cervical Cancer. Also, work extended to applied set of feature reduction techniques to reduce computation tasks and to improve model accuracy. However, in this cancer data total 4 target variables: Hinselmann, Schiller, Cytology, and Biopsy are considered associated with 32 risk factors. Moreover, the study used the number of benchmarks like Accuracy, Sensitivity, Specificity, Positive Prediction Accuracy (PPA) and Negative Prediction Accuracy (NPA) for the performance analysis. The results showed that the proposed ensemble classifier method to be proven efficient for cervical cancer analysis compared to standard methods.