COMPARATIVE STUDY OF TUMOR STAGE PREDICTION IN BREAST CANCER THROUGH FACTOR ANALYSIS AND MULTINOMIAL LOGISTIC REGRESSION

Abstract

Breast cancer remains one of the leading causes of cancer-related morbidity and mortality among women globally. Early-stage detection of tumor of breast cancer assists in providing preventive measures against its spread to other parts of the body. In this study, factor analysis was employed in reducing the number of genes to fewer principal components. The Scree and Eigenvalue methods of selecting the number of principal components were employed for comparison purpose. Multinomial logistic regression model was employed to fit the stage of tumor of the breast cancer on the scores of the principal component variables along with the patient’s age and tumor size. The findings of the study showed that the eigenvalue approach outperformed the Scree approach ranging from the percentage of variance explained to the accuracy level. Hence, the eigenvalue method of selecting number of components to include in factor analysis is recommended.