Comparison of Classification Models for Early Prediction of Breast Cancer

Abstract

Breast cancer is the second most leading cause of women's death in America. To create an accurate prediction model and analyze the remarkable risk factors, a data mining classification task that involves different methods has applied. Data mining has been used to extract hidden knowledge in different domains such as business, medicine, science, engineering, etc. This research aims to predict breast cancer using anthropometric data and parameters that are collected in routine blood analysis. First, we found the most important attributes in the dataset that can be selected as a Biomarker; by applying the recursive feature elimination method. We found that Age, BMI, Glucose, HOMA, and Resistin can be selected as the best Biomarker for breast cancer. We applied different classification techniques; K-NN, ANN, Decision trees, Naive Bayesian and found that artificial neural networks best classify the attribute with an accuracy of 80.00%. This study will also helps doctors and medical practitioners for early diagnosis of breast cancer.