Customer Churn Prediction In B2b Software As A Service

Abstract

Customer Retention Has Become A Major Concern For Subscription-based B2B Software As A Service (SaaS) Companies Because Long-term Contracts Directly Influence Revenue Stability. Even A Small Increase In Churn Rate Can Result In Noticeable Financial Loss For Service Providers. In This Study, Machine Learning Techniques Are Applied To Analyze And Predict Customer Churn Using Structured Enterprise Data. The Proposed Framework Includes Data Preprocessing, Handling Class Imbalance Using The Synthetic Minority Oversampling Technique (SMOTE), And Evaluating Several Ensemble Learning Models Including Random Forest, AdaBoost, CatBoost, And LightGBM. The Models Are Assessed Using Commonly Used Classification Metrics Such As Accuracy, Precision, Recall, F1-score, And ROC-AUC. Special Attention Is Given To Recall Since Identifying Potential Churn Customers Is Particularly Important For Business Decision-making. Among The Evaluated Approaches, LightGBM Demonstrated Stable And Comparatively Better Performance Across Multiple Metrics. To Improve Transparency, SHAP Analysis Is Used To Interpret Feature Contributions And Identify Factors Influencing Churn Behavior. The Trained Model Is Further Integrated Into A Streamlit-based Dashboard That Allows Real-time Predictions And Batch Processing Of Customer Data, Helping Organizations Monitor Churn Risk And Plan Proactive Retention Strategies.

Keywords