Methodology Pipeline
The complete workflow includes Data Preprocessing, SMOTEENN Balancing, Base Learner Training, and Stacked Ensemble Prediction.
Figure 1: The Indented Methodology Pipeline [Source: Ibad et al., 2025]
A Stacked Ensemble Model for Churn Prediction achieving 98.1% accuracy using the novel Latency Aware Accuracy Index (LAAI).
Customer churn is a critical challenge in the telecommunications industry. Traditional machine learning models often struggle with high-dimensional datasets and class imbalance. This research introduces a Stacked Ensemble Model that integrates Random Forest, KNN, and Naïve Bayes as base learners, with Logistic Regression as a meta-learner. By utilizing SMOTEENN for data balancing and minimizing the Latency Aware Accuracy Index (LAAI), the proposed model achieves superior performance suitable for real-time deployment.
The complete workflow includes Data Preprocessing, SMOTEENN Balancing, Base Learner Training, and Stacked Ensemble Prediction.
Figure 1: The Indented Methodology Pipeline [Source: Ibad et al., 2025]
We benchmarked our Stacking Classifier against industry-standard models. The ensemble approach not only achieved the highest accuracy but also maintained a high LAAI score, proving its efficiency.
High accuracy is useless if the model is too slow for real-time systems. We introduced the Latency Aware Accuracy Index to penalize slow models.