A study of feature selection techniques for predicting customer retention in telecommunication sector
by E. Sivasankar; J. Vijaya
International Journal of Business Information Systems (IJBIS), Vol. 31, No. 1, 2019

Abstract: Feature selection is the process of eliminating irrelevant features from the dataset, while maintaining acceptable classification accuracy. The selected features play an important role which can directly influence the effectiveness of the resulting classification. In this paper, a methodology is proposed consisting of two phases, attributes selection and classification based on the attributes selected. Phase one uses a filter and wrapper method for attribute selection with random over-sampling (Ros) through which the size of attributes set and misclassification error can be reduced. In the second phase, the selected attributes are taken as inputs by classification techniques like decision trees (DT), K-nearest neighbour (KNN), support vector machine (SVM), naive Bayes (NB) and artificial neural network (ANN). Finally, true churn, false churn, specificity and accuracy are measured to evaluate the efficiency of the proposed system and it is found that the above mentioned methodology performs well ahead for churn prediction and suits well for the telecommunication sector.

Online publication date: Wed, 08-May-2019

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Business Information Systems (IJBIS):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com