High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

An Efficient Statistical Based Approach for Outliers Detection

Author(s):

Pragya Soni , Lord Krishna College of Technology Indore ; Kamlesh Patidar, Lord Krishna College of Technology Indore

Keywords:

Databases, Machine learning, Data Mining

Abstract

Outliers are suspicious values because they are smaller or larger than the given values. The detection of outliers is important task in interest of data mining that realization that outliers as the key discovery from very large databases. Outlier’s detection is important because of various reasons such as human faults, system error or instrument error etc. Outlier detection techniques are divided into two categories: supervised and unsupervised. Supervised technique assumes the availability of training data set for normal as well as anomaly. Unsupervised technique does not require training data. This approach takes as input a set of data and finds outlier within the data. Sometimes good outliers give useful information for the discovery of new knowledge. Bad outliers are noisy data point. In this paper we propose a novel approach for outlier analysis along with review of some existing outlier detection techniques. In this paper we used some statistical properties to find out outlier. We also find that what the effects are when we consider analysis of data with outlier and without outliers. In this paper we proposed a novel approach in which we used some mathematical relationship between data after deleting the outliers.

Other Details

Paper ID: IJSRDV8I120002
Published in: Volume : 8, Issue : 12
Publication Date: 01/03/2021
Page(s): 50-52

Article Preview

Download Article