Welcome to the InfoSci Platform
Predicting Churn of Credit Card Customers Using Machine Learning and AutoML

Predicting Churn of Credit Card Customers Using Machine Learning and AutoML

Rajeev Kumar Gupta, Santosh Bharti, Nikhlesh Pathik, Ashutosh Sharma
Copyright: © 2022 |Volume: 13 |Issue: 3 |Pages: 19
ISSN: 1938-0232|EISSN: 1938-0240|EISBN13: 9781683180500|DOI: 10.4018/IJITPM.313422
Cite Article Cite Article

MLA

Gupta, Rajeev Kumar, et al. "Predicting Churn of Credit Card Customers Using Machine Learning and AutoML." IJITPM vol.13, no.3 2022: pp.1-19. http://doi.org/10.4018/IJITPM.313422

APA

Gupta, R. K., Bharti, S., Pathik, N., & Sharma, A. (2022). Predicting Churn of Credit Card Customers Using Machine Learning and AutoML. International Journal of Information Technology Project Management (IJITPM), 13(3), 1-19. http://doi.org/10.4018/IJITPM.313422

Chicago

Gupta, Rajeev Kumar, et al. "Predicting Churn of Credit Card Customers Using Machine Learning and AutoML," International Journal of Information Technology Project Management (IJITPM) 13, no.3: 1-19. http://doi.org/10.4018/IJITPM.313422

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

Nowadays, a major concern for most retail banks is the risk that originates from customer fluctuation and that increases the cost of almost every financial product. In this work, the authors compared different approaches and algorithms to predict the relevant features that affect the customer churn, which means we can find ways to reduce the customer churn and create financial inclusion. This research was conducted by applying different machine learning techniques like decision tree classifier, random forest classifier, AdaBoost classifier, extreme gradient boosting, and balancing data with random under-sampling and random oversampling. The authors have also implemented AutoML to further compare different models and improve the accuracy of the model to predict customer churn. It was observed that applying AutoML highest accuracy model gave the accuracy of 97.53% in comparison to that of the decision tree classifier, which was 93.48% with the use of low processing power. Important features were ‘total transaction amount' and ‘total transaction count' to predict customer churn for a given dataset.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.