{"title":"Customer Churn Prediction Using Four Machine Learning Algorithms Integrating Feature Selection and Normalization in the Telecom Sector","authors":"Alanoud Moraya Aldalan, Abdulaziz Almaleh","volume":195,"journal":"International Journal of Electronics and Communication Engineering","pagesStart":76,"pagesEnd":84,"ISSN":"1307-6892","URL":"https:\/\/publications.waset.org\/pdf\/10012996","abstract":"
A crucial part of maintaining a customer-oriented business in the telecommunications industry is understanding the reasons and factors that lead to customer churn. Competition between telecom companies has greatly increased in recent years, which has made it more important to understand customers\u2019 needs in this strong market. For those who are looking to turn over their service providers, understanding their needs is especially important. Predictive churn is now a mandatory requirement for retaining customers in the telecommunications industry. Machine learning can be used to accomplish this. Churn Prediction has become a very important topic in terms of machine learning classification in the telecommunications industry. Understanding the factors of customer churn and how they behave is very important to building an effective churn prediction model. This paper aims to predict churn and identify factors of customers\u2019 churn based on their past service usage history. Aiming at this objective, the study makes use of feature selection, normalization, and feature engineering. Then, this study compared the performance of four different machine learning algorithms on the Orange dataset: Logistic Regression, Random Forest, Decision Tree, and Gradient Boosting. Evaluation of the performance was conducted by using the F1 score and ROC-AUC. Comparing the results of this study with existing models has proven to produce better results. The results showed the Gradients Boosting with feature selection technique outperformed in this study by achieving a 99% F1-score and 99% AUC, and all other experiments achieved good results as well.<\/p>","references":"[1] H. Chen, R. H. Chiang, and V. C. Storey, \u201cBusiness intelligence and\r\nanalytics: From big data to big impact,\u201d MIS quarterly, pp. 1165\u20131188,\r\n2012.\r\n[2] T. Landis and S. Philips, \u201cCustomer retention\r\nmarketing vs. customer acquisition marketing,\u201d Apr\r\n2022. [Online]. Available: https:\/\/www.outboundengine.com\/blog\/\r\ncustomer-retention-marketing-vs-customer-acquisition-marketing\/\r\n[3] S. Yadav, A. Jain, and D. Singh, \u201cEarly prediction of employee attrition\r\nusing data mining techniques,\u201d in 2018 IEEE 8th International Advance\r\nComputing Conference (IACC), 2018, pp. 349\u2013354.\r\n[4] B. J. Ng), \u201cClassification analysis on telco customer\r\nchurn,\u201d Nov 2021. [Online]. Available: https:\/\/medium.com\/swlh\/\r\nclassification-analysis-on-telco-customer-churn-a01599ad28d7\r\n[5] C.-P. Wei and I.-T. Chiu, \u201cTurning telecommunications call details\r\nto churn prediction: a data mining approach,\u201d Expert systems with\r\napplications, vol. 23, no. 2, pp. 103\u2013112, 2002.\r\n[6] K. Coussement and D. Van den Poel, \u201cChurn prediction in subscription\r\nservices: An application of support vector machines while comparing\r\ntwo parameter-selection techniques,\u201d Expert systems with applications,\r\nvol. 34, no. 1, pp. 313\u2013327, 2008.\r\n[7] H. Abbasimehr, M. Setak, and M. Tarokh, \u201cA neuro-fuzzy classifier\r\nfor customer churn prediction,\u201d International Journal of Computer\r\nApplications, vol. 19, no. 8, pp. 35\u201341, 2011.\r\n[8] H. Jain, A. Khunteta, and S. Srivastava, \u201cChurn prediction in\r\ntelecommunication using logistic regression and logit boost,\u201d Procedia\r\nComputer Science, vol. 167, pp. 101\u2013112, 2020.\r\n[9] P. Lalwani, M. K. Mishra, J. S. Chadha, and P. Sethi, \u201cCustomer churn\r\nprediction system: a machine learning approach,\u201d Computing, pp. 1\u201324,\r\n2021. [10] M. Karanovic, M. Popovac, S. Sladojevic, M. Arsenovic, and\r\nD. Stefanovic, \u201cTelecommunication services churn prediction-deep\r\nlearning approach,\u201d in 2018 26th Telecommunications Forum (TELFOR).\r\nIEEE, 2018, pp. 420\u2013425.\r\n[11] G. Esteves and J. Mendes-Moreira, \u201cChurn perdiction in the telecom\r\nbusiness,\u201d in 2016 Eleventh International Conference on Digital\r\nInformation Management (ICDIM). IEEE, 2016, pp. 254\u2013259.\r\n[12] H. Jain, A. Khunteta, and S. P. Shrivastav, \u201cTelecom churn\r\nprediction using seven machine learning experiments integrating features\r\nengineering and normalization,\u201d 2021.\r\n[13] P. Swetha, S. Usha, and S. Vijayanand, \u201cEvaluation of churn rate using\r\nmodified random forest technique in telecom industry,\u201d in 2018 3rd IEEE\r\nInternational Conference on Recent Trends in Electronics, Information\r\n& Communication Technology (RTEICT). IEEE, 2018, pp. 2492\u20132497.\r\n[14] O. company, \u201cCustomer signature for churn analysis - dataset\r\nby bob-wakefield,\u201d Sep 2019. [Online]. Available: https:\/\/data.world\/\r\nbob-wakefield\/call-center-data\r\n[15] C. MAWER, \u201cThe value of exploratory data analysis,\u201d\r\nSep 2017. [Online]. Available: https:\/\/www.svds.com\/\r\nvalue-exploratory-data-analysis\/\r\n[16] s. babu, \u201cEda for machine learning: Exploratory data analysis in python,\u201d\r\nApr 2021. [Online]. Available: https:\/\/www.analyticsvidhya.com\/blog\/\r\n2021\/04\/rapid-fire-eda-process-using-python-for-ml-implementation\/\r\n[17] A. Ultsch, \u201cEmergent self-organising feature maps used for prediction\r\nand prevention of churn in mobile phone markets,\u201d Journal of Targeting,\r\nMeasurement and Analysis for Marketing, vol. 10, no. 4, pp. 314\u2013324,\r\n2002.\r\n[18] G. Dougherty, Pattern recognition and classification: an introduction.\r\nSpringer Science & Business Media, 2012.\r\n[19] D. Singh and B. Singh, \u201cInvestigating the impact of data normalization\r\non classification performance,\u201d Applied Soft Computing, vol. 97, p.\r\n105524, 2020.\r\n[20] A. Al Marouf, M. K. Hasan, and H. Mahmud, \u201cComparative analysis\r\nof feature selection algorithms for computational personality prediction\r\nfrom social media,\u201d IEEE Transactions on Computational Social\r\nSystems, vol. 7, no. 3, pp. 587\u2013599, 2020.\r\n[21] V. R, \u201cFeature selection - correlation and p-value,\u201d\r\nApr 2022. [Online]. Available: https:\/\/towardsdatascience.com\/\r\nfeature-selection-correlation-and-p-value-da8921bfb3cf\r\n[22] K. Pearson, \u201cX. on the criterion that a given system of deviations from\r\nthe probable in the case of a correlated system of variables is such that it\r\ncan be reasonably supposed to have arisen from random sampling,\u201d The\r\nLondon, Edinburgh, and Dublin Philosophical Magazine and Journal of\r\nScience, vol. 50, no. 302, pp. 157\u2013175, 1900.\r\n[23] M. Nikulin, \u201cChi-squared test for normality,\u201d in Proceedings of\r\nthe International Vilnius Conference on Probability Theory and\r\nMathematical Statistics, vol. 2, no. 1, 1973, pp. 119\u2013122.\r\n[24] R. Bevans, \u201cAn introduction to t-tests,\u201d Dec 2020. [Online]. Available:\r\nhttps:\/\/www.scribbr.com\/statistics\/t-test\/\r\n[25] A. A, \u201cStatistical tests: Feature selection using statistical tests,\u201d Jun\r\n2021. [Online]. Available: https:\/\/www.analyticsvidhya.com\/blog\/2021\/\r\n06\/feature-selection-using-statistical-tests\/\r\n[26] J. Han, M. Kamber, and J. Pei, \u201cData mining: Concepts and techniques\r\nthird edition [m],\u201d The Morgan Kaufmann Series in Data Management\r\nSystems, vol. 5, no. 4, pp. 83\u2013124, 2011.\r\n[27] J. H. Friedman, \u201cGreedy function approximation: a gradient boosting\r\nmachine,\u201d Annals of statistics, pp. 1189\u20131232, 2001.\r\n[28] J. Hoare, \u201cGradient boosting explained - the coolest kid on the\r\nmachine learning block - displayr,\u201d https:\/\/www.displayr.com\/\r\ngradient-boosting-the-coolest-kid-on-the-machine-learning-block\/,\r\n(Accessed on 05\/05\/2022).\r\n[29] U. S\u00b8 . G\u00a8ursoy, \u201cCustomer churn analysis in telecommunication sector,\u201d\r\nI\u02d9stanbul U\u00a8 niversitesi I\u02d9s\u00b8letme Faku\u00a8ltesi Dergisi, vol. 39, no. 1, pp. 35\u201349,\r\n2010.\r\n[30] H. Sharma and S. Kumar, \u201cA survey on decision tree algorithms of\r\nclassification in data mining,\u201d International Journal of Science and\r\nResearch (IJSR), vol. 5, no. 4, pp. 2094\u20132097, 2016.\r\n[31] P. Yadav, \u201cDecision tree in machine learning \u2014 by prince\r\nyadav \u2014 towards data science,\u201d https:\/\/towardsdatascience.com\/\r\ndecision-tree-in-machine-learning-e380942a4c96, 11 2018, (Accessed\r\non 05\/05\/2022).\r\n[32] K. Singh, \u201cHow to dealing with imbalanced classes in machine\r\nlearning,\u201d Oct 2020. [Online]. Available: https:\/\/www.analyticsvidhya.\r\ncom\/blog\/2020\/10\/improve-class-imbalance-class-weights\/\r\n[33] H. He and Y. Ma, Imbalanced learning: foundations, algorithms, and\r\napplications. Wiley-IEEE Press, 2013.\r\n[34] J. Brownlee, \u201cRoc curves and precision-recall curves for\r\nimbalanced classification,\u201d https:\/\/machinelearningmastery.com\/\r\nroc-curves-and-precision-recall-curves-for-imbalanced-classification\/,\r\nJan 2020, (Accessed on 05\/05\/2022).","publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 195, 2023"}