Forecasting the survival rate of breast cancer patients using a supervised learning method

The paper aims to develop a regression model using the NKI breast cancer data set. The methodology used to achieve the objectives includes three variations of regression methods viz., linear, multiple, and polynomial, respectively. Regression analysis is one of the efficient predictive modeling methods that help understand the mathematical relationship between the variables. The multiple and polynomial regression methods also work in line with the linear regression model, but the number of independent variables will be varying. Queries related to health care data are of practical interest. The outcome of the predictive model helps in analyzing the behavior of different features of the breast cancer data set and provides useful insights towards the diagnosis of a patient. 14 out of 1570 useful features of the NKI data set are selected for the regression analysis. With different combinations of independent and dependent variables, it is found that multiple regression performs better with 83% accuracy.