Classification Modelling: A Case Study of Breast Cancer Patients of Islamabad

Abstract

Purpose: The rate of breast cancer in Pakistan is the highest among all other Asian countries and all other types of cancer. The foremost treatment for breast cancer patients of stage 2 and stage 3 is surgery. The main types of surgery in this era are Mastectomy and Breast Conservative surgery. The decision about the type of surgery depends on the demographic and clinical factors. Approach: In this study, the seven characteristics have been considered.  A purposive sample of 365 breast cancer patients were collected from the two main hospitals in Islamabad. The foremost objective of this study was to classify each breast cancer patient regarding surgery type based on significant explanatory characteristics. The binary logistics regression and discriminant analysis techniques were used and the significance of each parameter was tested. Findings: The main effects i.e., age, tumor size, Estrogen Receptor, and Progesterone Receptor were found to be significant with some diverse probabilities and all two-factor interactions were found to be non-significant. The sensitivity of logistic regression and discriminant analysis is almost the same i.e., 93.1% and 92.8% respectively whereas the specificity of these two techniques is also almost the same i.e., 70.8% and 71.9% respectively. The overall actual correct classify rate and Apparent error rate of both these techniques are found to be 87.7% and 12.3% respectively. Implications: In brief, it was deducted that the Tumor size stage is the most imperative characteristic among other significant characteristics in discriminating between two types of surgery