Stacking Ensemble Approach for Churn Prediction: Integrating CNN and Machine Learning Models with CatBoost Meta-Learner

n the telecom industry, predicting customer churn is crucial for improving retention. In literature, use of single classifiers predominantly focused. Customer data complex due to class imbalance and contain multiple factors that exhibit nonlinear dependencies. these scenarios, may be unable fully utilize available information capture underlying interactions effectively. contrast, ensemble learning combines various base empowers a more thorough analysis, leading improved prediction performance. this paper, heterogeneous model proposed in industry. The involves exploratory pre-processing resampling handle imbalance. model, trained with different characteristics are integrated through stacking technique. Specifically, convolutional-based neural network, logistic regression, decision tree Support Vector Machine (SVM) considered as work. utilizes unique strengths each classifier leverages collective knowledge improve performance meta-learner. efficacy assessed on real-world dataset, i.e., Cell2Cell. empirical results demonstrate superiority 62.4 % f1-score 60.62 recall.