THE RESEARCH AND ANALYSIS OF THE FUZZY K-NEAREST NEIGHBOR ALGORITHMS USING DIFFERENT METRICS FOR BREAST CANCER DIAGNOSIS

Cancer diagnosis is one of the most extensively discussed issues in medicine. A number of researchers focused on improving the work efficiency to obtain satisfactory results. The most dangerous cancer is breast cancer, hence a special attention given to its diagnostics. Machine learning is widely used in bioinformatics and particularly in diagnostics of breast cancer. In this article the fuzzy K-Nearest Neighbor algorithm is used, which is one of the most popular clustering methods. The purpose of this study is to research and analyze fuzzy K-Nearest Neighbor algorithms using different distance functions including Chebyshev distance, Euclidean distance, Minkowski distance and Manhattan distance. The quality of application results of the algorithm depends to a large extent on the choice of distance and value of "k" that represents the number of nearest neighbors. In this study, we examine the efficiency of different distances that are applicable for the fuzzy K-NN algorithm. Besides, we analyze the application results of different metrics using various values of "k" and several classification rules. Our study shall be performed on the basis of WBCD (Wisconsin Breast Cancer Database).