Outliers Analysis with Fuzzy Clustering Model

Outlier detection is important in many fields. In statistics, an outlier is a observation that is numerically far-away from the rest of the data. The handling of outlying observations in a data set is one of the most important tasks in data pre-processing. The large data base can be classified in an unsupervised manner using clustering and classification algorithms. Fuzzy C-means is a method of clustering which was developed by Dunn in (1973) and improved by Bezdek in (1981). This allocates one piece of data in two or more clusters and it is frequently used in pattern recognition. Herein a proposed method based on Fuzzy approach which combines outlier analysis and clustering technique is presented. Clustering validation technique adaptively evaluated the results of a clustering algorithm. A numerical example is provided for illustration using iris data set.