A Modification on K-Nearest Neighbor Classifier

K-Nearest Neighbor (KNN) classification is one of the most fundamental and simple classification methods. When there is little or no prior knowledge about the distribution of the data, the KNN method should be one of the first choices for classification. In this paper a modification is taken to improve the performance of KNN. The main idea is to use robust neighbors in training data. This modified KNN is better from traditional KNN in both terms: robustness and performance. The proposed KNN classification is called Modified K-Nearest Neighbor (MKNN). Inspired from the traditional KNN algorithm, the main idea is to classify an input query according to the most frequent tag in set of neighbor tags. MKNN can be considered a kind of weighted KNN so that the query label is approximated by weighting the neighbors of the query. The procedure computes the frequencies of the same labeled neighbors to the total number of neighbors. The proposed method is evaluated on a variety of several standard UCI data sets. Experiments show the excellent improvement in the performance of KNN method.