|
چکیده
|
Aim: Breast cancer is one of the most common cancers among women compared to all other ones. Proper and early detection of tumor type in a patient is a critical subject that can increase the survival chance. Statistical methods, such as classification can be useful instrument for physicians as an auxiliary tool to detect the type of tumor, increase the speed, reduce the cost and time of diagnosis as well as reduce the error of physician in the diagnosis of the tumor. Methods: In this paper, we use multivariate linear regression, logistic regression, the K-nearest neighbor (KNN) method and discriminant analysis to determine tumor type in a patient using Wisconsin Breast Cancer (WBC) database. In addition to linear methods, quadratic discriminant analysis will be used. Stepwise method for variable selection is used in regression method. Results: The accurate percentage of classification on testing data, depends on selected method and the number of independent variables and at least equal to 90.8% for logistic regression and a maximum 99.6% for KKN with K=9. Our results show about 3.1% of doctors’ diagnoses in the breast cancer may be incorrect. Conclusion: In General, if the assumptions used in each method is established, it can be said that the difference is not significant in choosing model type, and all of these methods are good tool for increasing precision of breast diagnosis.
|