Analysis Of Risk Identification For Churn Prediction In Telecom Industry, PT. Telkom Indonesia Using Machine Learning Technique Xgboost Algorithm By Customer Behaviour

Eben, Jahfal Hizbullah Putra (2025) Analysis Of Risk Identification For Churn Prediction In Telecom Industry, PT. Telkom Indonesia Using Machine Learning Technique Xgboost Algorithm By Customer Behaviour. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026211195-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Prediksi churn merupakan komponen penting dalam Manajemen Hubungan Pelanggan
(CRM) di industri telekomunikasi, yang memungkinkan bisnis untuk menerapkan strategi
retensi yang efektif dan mengoptimalkan upaya pemasaran untuk mengurangi churn pelanggan.
Namun, menangani ketidakseimbangan kelas, yang merupakan masalah umum dalam dataset
churn, tetap menjadi tantangan. Studi ini mengusulkan pendekatan prediksi churn
menggunakan XGBoost, sebuah metode ensemble learning yang dikenal karena kekuatan dan
skalabilitasnya, yang dipadukan dengan Teknik Oversampling Minoritas Sintetis (SMOTE)
untuk menangani data yang tidak seimbang. Eksperimen komprehensif pada dataset benchmark, termasuk IBM Telco dan Cell2Cell, menunjukkan bahwa XGBoost, yang dipadukan dengan
SMOTE, mencapai kinerja prediktif yang lebih baik dibandingkan dengan model tradisional, unggul dalam akurasi, presisi, recall, dan skor F1. Temuan ini menyoroti efektivitas XGBoost dalam mengatasi tantangan ketidakseimbangan kelas dan memberikan solusi yang menjanjikan untuk prediksi churn di sektor telekomunikasi.
========================================================================================================================================
Churn prediction is a critical component of Customer Relationship Management (CRM) in the telecommunications industry, enabling businesses to implement effective retention strategies and optimize marketing efforts to reduce customer attrition. However, addressing class imbalance, a common issue in churn datasets, remains a challenge. This study proposes a churn prediction approach using XGBoost, an ensemble learning method known for its robustness and scalability, combined with the Synthetic Minority Oversampling Technique (SMOTE) to handle imbalanced data. Comprehensive experiments on benchmark datasets, including IBM Telco and Cell2Cell, demonstrate that XGBoost, coupled with SMOTE, achieves superior predictive performance compared to traditional models, excelling in accuracy, precision, recall, and F1-score. These findings highlight the effectiveness of XGBoost in overcoming class imbalance challenges and provide a promising solution for churn prediction in the telco sector.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	XGBoost, Telecommunication, Churn, Smote, Predictive, XGBoost, Telekomunikasi,Pemutus Layanan, Smote, Prediksi
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Jahfal Hizbullah Putra Eben
Date Deposited:	28 Jul 2025 08:33
Last Modified:	28 Jul 2025 09:16
URI:	http://repository.its.ac.id/id/eprint/122555

Actions (login required)

View Item