Explainable Customer Churn Prediction in Telecom Using Ensemble Learning and SHAP Analysis
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Predicting customer churn is crucial for telecommunications companies, as retaining existing customers is more cost-effective than acquiring new ones.This work proposes a novel Stacking ensemble framework integrating five base classifiers: Decision Tree, Random Forest, Extra Trees, Gradient Boosting, and XGBoost, designed to accurately predict churn while providing interpretable explanations of model decisions.The methodology involves comprehensive data preprocessing, including outlier detection, handling of high-cardinality categorical variables, normalization and application of Synthetic Minority Over-sampling Technique (SMOTE), a technique to construct the synthetic samples of the minority group to overcome the class imbalance on a training set of 3,333 samples.Ensemble methods such as Soft Voting, Hard Voting, and the proposed Stacking approach are evaluated, with the Stacking ensemble achieving superior performance 94.75% accuracy, 73.20% recall, 88.75% precision, and an F1score of 80.23%.This represents a 3.09% improvement over the best previously reported accuracy of 91.66% and outperforms individual models, including XGBoost (F1-score 79.14%).Model interpretability is enhanced through Shapley additive explanations (SHAP), highlighting total day minutes, international plan subscription, and account length as key predictors influencing churn.The proposed framework offers a reliable and transparent tool for churn prediction applicable in business contexts requiring explainable AI.Future work will explore integrating temporal deep learning models and real-time updated data to further improve predictive performance across diverse industries.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it