Comparative Analysis with Implementation of Cluster Based, Distance Based and Density Based Outlier Detection Techniques Using Different Healthcare Datasets

Abstract

Outliers is view as an error data in information which is turned into important crisis that has been investigated in various areas of study plus functional fields. Several outlier detection methods have been implemented to assured functional fields, whereas several methods are supplementary basic. Various functional areas are also investigated in severe privacy like study on offense as well as terrorist behaviors. Through the improvement in information skills, the numeral of records, plus their measurement as well as difficulty, raise fast, that outcome in the need of computerized examination of huge quantity of various ordered data. For this intention, different data mining systems are utilized. The objective of these types of systems is to detect unseen dependencies from the records. Outlier detection in data mining is the detection of objects, remarks or observations that doesn’t match to a predictable sample in a set of record. This detection technique is more beneficial in the several areas such as health trade, offense finding, fake operation, community protection and so on. In this paper we have studied different outlier detection algorithms such as Cluster based outlier detection, Distance based outlier detection plus Density based outlier detection. Result experimentation is done on different four dataset to identify the outliers and the comparative result shows that the cluster based methods are efficient for calculation of clusters and density-based outlier detection algorithm offers improved accuracy and faster execution for identification of outliers than other two outlier detection algorithm.