A Strategic Framework for Feature Selection in Banking Sector for Credit Risk Analysis
Femina Bahari T1, Sudheep Elayidom M2
1Femina Bahari T, Department of Computer Science & Engineering, Cochin University of Science and Technology, Kerala, India.
2Sudheep Elayidom M, Department of Computer Science & Engineering, Cochin University of Science and Technology, Kerala, India.
Manuscript received on 06 April 2019 | Revised Manuscript received on 10 May 2019 | Manuscript published on 30 May 2019 | PP: 1776-1780 | Volume-8 Issue-1, May 2019 | Retrieval Number: A1865058119/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Feature selection in data mining is critical as it generates optimum subset of features which are relevant for any classification model study. Selecting the optimum subset of features helps to improve the performance of the classifier. A large volume of data gets accumulated on a daily basis in banking industry as part of its various operations. Data is collected and stored in data warehouses for further processing. Information obtained from this data related to customers, transactions, services etc, if analyzed closely can contribute to the growth of industry to a large extent. But the vastness of data and the large number of features in the database makes this analysis a tedious process. Feature selection helps to remove the irrelevant features that adversely affect the performance of learning process in a classifier. A framework combining both filter and wrapper approaches are proposed in our work and an aggregate ranking strategy is followed to select the best features for classification. Naïve Bayes classifier is used in the wrapper method. A district bank dataset is used for testing both approaches and optimum features are selected for further experimental studies in classification.
Keywords: Credit Risk, Feature Selection, Filter Approach, Wrapper Method, Naïve Bayes Classifier.
Scope of the Article: Patterns and Frameworks