ISSN 2630-0583 (Print)

ISSN 2630-0656 (Online)


Journal of Current Science and Technology

Rangsit Journal of Arts and Sciences. Vol.2 No.1 , January - June 2012.

Multinomial logistic regression analysis of breast cancer

Doungporn Maiprasert and Krieng Kitbumrungrat


This study aims at developing Multinomial Logistic Regression (MLR) to evaluate the probability of breast cancer, proposing MLR to predict five stages of breast cancer (Benign, I, II, III and IV).  Nine characteristics of breast cancer: Clump Thickness (X1); Uniformity of Cell Size (X2); Uniformity of Cell Shape (X3); Marginal Adhesion (X4); Single Epithelial Cell Size (X5); Bare Nuclei (X6); Bland Chromatin (X7); Normal Nucleoli (X8); and Mitoses (X9) are used as  independent variables.  Results show that Multinomial Logistic Regression (MLR) yields a coefficient of a model indicating that X1 and X6 have significance less than 0.05. Thus, the prediction of log – likelihood function for a classification staging of breast cancer with P(Y<4) of stage IV is a reference category, reducing a model as: log(Pi/1-Pi) = 819.332 + 608.852x1 + 615.165x6

Keywords: multinomial logistic regression(MLR), logistic regression, classification, prediction

Download Full Paper.