Classification Based Analysis on Cancer Datasets Using Predictor Measures

Abstract

Cancer is a life-threatening disease. Probably the most effective way to reduce cancer deaths is to detect it earlier. Diagnosing the disease earlier needs an accurate and reliable procedure which could be used by physicians to distinguish between cancer from malignant ones without leaving for surgical biopsy. Data mining offers solution for such types of the problems where a large quantity of information about patients and their conditions are stored in clinical database. This paper focuses on prediction of some such diseases like Leukemia and Breast cancers. Naïve Bayes and SVM prediction models are built for the prediction and classification. The performance of the proposed models produced significant results of above 96% while compared with other models in terms of accuracy, computational time and convergence.

Keywords: Prediction, Data Mining, Diagnosis, Cancer, Naïve Bayes, Supper Vector machine (SVM).

DOI: 10.7176/CEIS/10-6-05

Publication date:July 31^st 2019