Abstract

AbstractDetecting Breast Cancer in the early step is significant to minimize the death rate by enhancing the available treatments. As there exists no definitive treatment for Breast Cancer detection, early detection is significant. Hence, this study employs Machine Learning (ML) methods for expecting Breast Cancer effectively with efficiently. In this study, novelty is given in the feature selection and classification process. This study intends to perform relevant feature selection through the proposed Greedy Optimization (GO) method. In addition, it aims to perform a classification of the relevant features by the use of the newly introduced Enlarge C4.5 algorithm (Ext.-C4.5). At first, Breast Cancer dataset is loaded. Then, pre-processing is performed. The pre-processed data is obtained through data reduction and dimensionality reduction. Here, data reduction is performed using the Block Level Deduplication, dimensionality reduction is performed using C-Isomap. This pre-processed data is considered for feature selection. This feature selection is given for train and test split. This is followed by classification. Lastly, the prediction is performed by the trained model. The experimental implementation and performance analysis of the planned system is undertaken. It is performed by comparing the proposed system with the existing systems (Naïve Bayes (NB), Decision Tree (DT), Nearest Neighbour (NN), Multilayer Perceptron (MLP), Artificial Neural Network (ANN), Extreme Learning Machine, K-Nearest Neighbour (KNN) and Support Vector Machine (SVM)) in terms of correctness, accuracy, specificity, sensitivity, f1-score and Completing Time. The results revealed that the proposed GO and E-C4.5 performs efficiently than the existing methodology.KeywordsBreast cancer predictionBlock level deduplicationC-IsomapGreedy optimizationEnlarge C4.5 algorithm

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call