Analysis of Distance Measures Using K-Nearest Neighbor Algorithm on KDD Dataset

Classification is the process of analyzing the input data and building a model that describes data classes. K-Nearest Neighbor is a classification algorithm that is used to find class label of unknown tuples. Distance measure functions are very important for calculating distance between test and training tuples. Main aim of this paper is to analyze and compare Euclidian distance, Chebychev distance and Manhattan distance function using K-Nearest Neighbor. These distance measures are compared in terms of accuracy, specificity, sensitivity, false positive rate and false negative rate on KDD dataset. Manhattan distance gives high performance.