Ensemble feature subset selection : integration of symmetric uncertainty and chi-square techniques with RreliefF
By: Sumant, Archana Shivdas
.
Contributor(s): Patil, Dipak.
Publisher: New York Springer 2022Edition: Vol.103(3), June.Description: 831-844p.Subject(s): Humanities and Applied SciencesOnline resources: Click here
In:
Journal of the institution of engineers (India): Series BSummary: The emanation of the high-dimensional data processing induces severe problems and challenges besides the apparent benefits. High-dimensional data analysis demands a huge requirement for processing. In this paper, we have proposed multistage methods ChS-R (Chi-square integrated with RReliefF) and SU-R (Symmetric Uncertainty integrated with RReliefF) for ranking features. The proposed integrated feature ranking methods use different statistical methods to select appropriate feature subset. The methods are integrated to overcome issues of one method with benefits of other method. The Chi-square (ChS) test is initially applied to select top n features, followed by RReliefF. In RReliefF algorithm, attributes are selected according to their suitability for the target function. It gives global view of attribute quality for further dimensionality reduction. In addition RReliefF deals with noisy, incomplete and multi-class data. Similarly, Symmetric Uncertainty (SU) integrated with RReliefF approach is proposed. The results are validated with random forest (RF), K-nearest neighbor (KNN), support vector machine (SVM) classifiers. The proposed systems are compared with SU, ChS, Relief and Ensemble Feature Selection with Mutual Information (EFS-MI) methods. The proposed approach achieves 89.48% dimensionality reduction.
| Item type | Current location | Call number | Status | Date due | Barcode | Item holds |
|---|---|---|---|---|---|---|
Articles Abstract Database
|
School of Engineering & Technology Archieval Section | Not for loan | 2022-1299 |
The emanation of the high-dimensional data processing induces severe problems and challenges besides the apparent benefits. High-dimensional data analysis demands a huge requirement for processing. In this paper, we have proposed multistage methods ChS-R (Chi-square integrated with RReliefF) and SU-R (Symmetric Uncertainty integrated with RReliefF) for ranking features. The proposed integrated feature ranking methods use different statistical methods to select appropriate feature subset. The methods are integrated to overcome issues of one method with benefits of other method. The Chi-square (ChS) test is initially applied to select top n features, followed by RReliefF. In RReliefF algorithm, attributes are selected according to their suitability for the target function. It gives global view of attribute quality for further dimensionality reduction. In addition RReliefF deals with noisy, incomplete and multi-class data. Similarly, Symmetric Uncertainty (SU) integrated with RReliefF approach is proposed. The results are validated with random forest (RF), K-nearest neighbor (KNN), support vector machine (SVM) classifiers. The proposed systems are compared with SU, ChS, Relief and Ensemble Feature Selection with Mutual Information (EFS-MI) methods. The proposed approach achieves 89.48% dimensionality reduction.
Click on an image to view it in the image viewer
There are no comments for this item.