 |
International Journal of Computer Science and Research
ISSN : 2210-9668
|
|
|
|
Abstract
| Title |
: |
Selection of Best Outlier Detection Method Using Regression Analysis |
| Authors |
: |
Fahad Sultan, Mudassir Ahmed |
| Keywords |
: |
Outlier; Outlier Detection; Regression; Analysis; Data; Applications |
| Issue Date |
: |
September 2010 |
| Abstract |
: |
Outliers are unusual data values that are inconsistent with most of the records. Such non-representative records can seriously affect the model to be produced, so detecting outlier is a significant job to achieve higher accuracy. Several outlier detection methods are used in literature for real as well as simulated data sets. The aim of this study is to compare the two outlier detection method i.e. Cook’s Distance and Mahalnobis method with standard method for outlier detection. The 15 replicates for simple linear regression of total household expenditure and household size are used to find the most efficient outlier detection method. It is found that the standard method of outlier detection produce less residual to predict the actual values as compared to the other two methods. While among the other two methods Cook’s method produce less prediction error as compared to the Mahalanobis methods. |
| Page(s) |
: |
1-9 |
| ISSN |
: |
2210-9668 |
| Source |
: |
Vol. 1, No.1 |
|
|
|
|
|