A Bonferroni Mean Based Fuzzy K Nearest Centroid Neighbor Classifier

K-nearest neighbor (KNN) is an effective nonparametric classifier that determines the neighbors of a point based only on distance proximity. The classification performance KNN disadvantaged by presence outliers in small sample size datasets and its deteriorates with class imbalance. We propose local Bonferroni Mean Fuzzy K-Nearest Centroid Neighbor (BM-FKNCN) assigns label query dependent nearest centroid mean vector to better represent underlying statistic dataset. proposed robust towards because Nearest Neighborhood (NCN) concept also considers spatial distribution symmetrical placement neighbors. Also, can overcome domination imbalance it averages all vectors from each adequately interpret classes. BM-FKNCN tested Knowledge Extraction Evolutionary Learning (KEEL) repository benchmarked results KNN, Fuzzy-KNN (FKNN), BM-FKNN FKNCN classifiers. experimental show achieves highest overall average accuracy 89.86% compared other four