Recognisation of Outlier using Distance based method for Large Scale Database

Abstract

This paper studies the difficulties of outlier detection on inexact data. We study the normal instances for each uncertain object using the instances of objects with analogous properties. Outlier detection is a significant research problem in data mining that goals to determine valuable abnormal and irregular patterns hidden in vast data sets. Most existing outlier detection approaches only deal with static data with comparatively low dimensionality. Newly, outlier detection for high-dimensional stream data turn into a new emergent research problem. A key remark that inspires this research is that outliers in high-dimensional data are predictable outliers, i.e., they are embedded in lower dimensional subspaces. Detecting projected outliers from high-dimensional stream data is a very stimulating task for numerous reasons. The paper shows the detailed study of outlier detection algorithms and its results also.