Feature Based Ensemble Learning Model for Breast Cancer Reoccurrence Retrieval

In recent years, 30% of women have been diagnosed with cancer yearly. Improvement of medical treatments made patients would be in remission but with challenges. It's estimated that there will be 14.8% of new cases in the last year. Breast cancer reoccurrence regenerates new challenges causing severe effects and even causing life loss. So, if it is detected early, it can cure it. Various latest techniques like machine learning are very much required for solving and predicting the reoccurrence for reducing mortality to some extent. The research paper proposed the ensemble approach, which used the Voting method for combining techniques used ensemble methods for detecting the two classes of tumors benign and malignant. Ensemble used Meta to implement more than one classifier. Experiments conducted on Wisconsin Diagnostic Breast Cancer (WDBC) dataset and voting techniques are used to get better results for model evaluation. Logistics regression, support vector machine, RBFF and Linear classifier, decision tree classifier and Random Forest are used to get classification accuracy, precision, recall, and F1 measure. The results obtained show that the ensemble models showed significant achievement in terms of performance and 98 % in terms of accuracy. With AdaBoost and cost-sensitivity in a model, a reasonable accuracy has been achieved. The proposed model in this research supports setting and evaluating various follow-up visit interventions and very advanced treatment recommendations, so there should be very low cancer mortality.