PREDIKSI DAN ANALISIS CUSTOMER CHURN
PADA PERUSAHAAN TELKOMSEL
DENGAN PENDEKATAN MACHINE LEARNING
Prediction and Analysis of Customer Churn at Telkomsel Using Machine Learning Approach
Customer churn menjadi salah satu permasalahan utama dalam industri telekomunikasi, termasuk di Telkomsel, yang merupakan operator seluler terbesar di Indonesia. Penelitian ini bertujuan untuk membangun model klasifikasi guna memprediksi churn pelanggan serta menganalisis faktor-faktor yang memengaruhi churn menggunakan pendekatan CRISP-DM. Data diperoleh melalui kuesioner online dari 100 responden yang merupakan mahasiswa aktif Universitas Negeri Surabaya. Proses penelitian mencakup tahapan data preparation (normalisasi, encoding, dan penghapusan atribut tidak relevan) serta penerapan algoritma klasifikasi seperti Decision Tree, Random Forest, K-Nearest Neighbors, Support Vector Machine, dan Naïve Bayes. Evaluasi dilakukan menggunakan metrik seperti akurasi, presisi, recall, dan F1-Score. Hasil penelitian menunjukkan bahwa Random Forest merupakan algoritma terbaik dengan nilai F1-Score sebesar 87.50% pada rasio data 80:20. Analisis fitur menunjukkan bahwa atribut status churn sebelumnya memiliki pengaruh terbesar terhadap prediksi churn.
Kata Kunci: Customer Churn, Komparasi Algoritma, Telkomsel, Klasifikasi, CRISP-DM.
Customer churn is one of the main challenges in the telecommunications industry, including Telkomsel, the largest mobile operator in Indonesia. This study aims to develop a classification model to predict customer churn and analyze the factors influencing churn using the CRISP-DM methodology. Data were collected through an online questionnaire from 100 respondents, consisting of active students at Universitas Negeri Surabaya. The research process involved data preparation (normalization, encoding, and removal of irrelevant attributes) and the application of classification algorithms such as Decision Tree, Random Forest, K-Nearest Neighbors, Support Vector Machine, and Naïve Bayes. Evaluation metrics included accuracy, precision, recall, and F1-Score. The results show that Random Forest is the best-performing algorithm with an F1-Score of 87.50% using an 80:20 data split. Feature analysis revealed that the prior churn status attribute has the most significant impact on churn prediction.
Keywords: Customer Churn, Algorithm Comparison, Telkomsel, Classification, CRISP-DM