Tri-Strategy Remora Optimization Algorithm based Support Vector Machine for Customer Churn Prediction

In the Telecommunication Industry (TCI), Customer Churn (CC) is an essential problem because to revenue of service providers is completely based on retaining current customers. Therefore, an effective Churn Prediction (CP) method is needed to monitor CC. However, numerous existing methods are there to predict CC, but they have drawbacks like null relevant features are selected to predict CC. This proposed a Tri-Strategy Remora Optimization Algorithm (TS-ROA) based Support Vector Machine (SVM) for predicting the churn customer. There are two datasets used for evaluating the proposed method are IBM Telco dataset and the CCP 2020 dataset. The data are pre-processed by label encoding, normalization, and balancing the dataset. After pre-processing, the feature selection process is performed by the proposed TS-ROA technique and at last prediction is done by SVM. The proposed TS-ROA-based SVM prediction method reached 97.36% accuracy, 96.71% precision, 96.33% recall and 96.59% f1-score on IBM Telco dataset and reached 96.82% accuracy, 96.51% precision, 96.28% recall and 96.39% f1-score on CCP dataset. The proposed prediction method shows increased performance on two datasets when compared to existing methods like ChurnNet and Grey Wolf Optimizer and Ensemble Neural Network (GWO-ENN).