Skripsi

KNOWLEDGE DISCOVERY PADA CHURN PELANGGAN E-COMMERCE MENGGUNAKAN INTERPRETABLE MACHINE LEARNING: STUDI KOMPARATIF MODEL BERBASIS SHAP

---

Customer churn remains a major challenge in the e-commerce industry as it directly impacts revenue and long-term customer value. This study applies an interpretable machine learning approach to not only predict churn but also identify the key factors influencing it. The analysis uses a publicly available dataset containing customer behavior and transaction records, with data preparation steps including handling of missing values, label encoding, and class balancing using SMOTE. Five classification models Logistic Regression, Random Forest, XGBoost, Support Vector Machine, and Gradient Boosting were evaluated based on accuracy, precision, recall, F1-score, and AUC. The results show that XGBoost achieved the best performance with 96% accuracy and an AUC of 0.999, followed by Random Forest. The interpretability analysis using SHAP revealed Tenure, Complain, and CashbackAmount as the most influential predictors. Longer customer relationships were associated with lower churn risk, while frequent complaints and high cashback usage increased the likelihood of churn. This study contributes by providing a predictive model that combines high accuracy with interpretability, enabling e-commerce businesses to design more effective and data-driven customer retention strategies.

---

Availability

Detail Information

Series Title

-

Call Number

T1894542025

Publisher

Indralaya : Prodi Sistem Informasi, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xvi, 95 hlm.; ilus.; tab, 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.310 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Informasi
Pembelajaran Mesin--Interpretable Machine Learning

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment