Detecting Outliers

This report presents an integrated outlier detection method, which is named "An Approach to Detect Outlier by Integrating Univariate Outlier Detection and K-means Algorithm". It provides efficient outlier detection and data clustering capabilities in the presence of outliers, and based on filtering of the data after univariate analysis. This algorithm is divided into two stages. The first stage provides Univariate outlier analysis. The main objective of the second stage is an iterative removal of objects, which are far away from their cluster centroids by applying K-means algorithm. The removal occurs according to the minimisation of the value of sum of the distances of all the points to their respective centroid in all the clusters. Finally, we provide experimental results from the application of our algorithm on several datasets to show its effectiveness and usefulness. The empirical results indicate that the proposed method was successful in detecting outliers and promising in practice.