An Automated Ensemble-Based Classification Model for The Early Diagnosis of The Cancer Using a Machine Learning Approach

Abstract

With the rapid growing of population, there is a need for a robust Computer Aided Diagnostic (CAD) system for health monitoring process. This will enable the domain specialist to have higher efficiency and throughput thus making the diagnostic part of health care more reliable & affordable. Hitherto pattern classification methods were widely used in developing such systems. Such system suffers from various drawbacks such as instability in recognition rate, unreliable performance over unknown instances and inability to deal with imbalanced data. Hence it is better to utilize the more advanced methods from Machine Learning techniques which incrementally learns the system in such situations. Predicting the radiological characteristic rating is one such typical problem which poses many challenging issues such as huge dimensionality, class imbalance and unavailability of samples for many classes. Hence, in this paper, proposed a Markov Bagging Ensemble Classifier (MBEC-DECORATE) for the automated classification. The proposed model uses the conventional ensemble classifier integrated with the Markov model. The performance expressed that proposed MBEC model exhibits higher accuracy of 99% for the J48 and random tree classifier model.