Using Big Data And Explainable Ai In A Hybrid Model To Predict Churn And Boost Customer Retention In Streaming Services

Abstract

For streaming services, where keeping current 
members is vital for ongoing success, customer 
churn prediction is a major concern.  This paper 
offers a large data-driven hybrid model predicting 
customer turnover with excellent accuracy by 
sophisticated machine learning and deep learning 
technologies.  The version addresses data imbalance 
using SMOTE oversampling on the Churn data 
dataset.  Predictive performance is improved by 
means of Chi-square (Chi2) and Sequential feature 
selection (SFS), optimising feature selection.  
Although several algorithms were used, the 
emphasis is on a voting Classifier combining 
boosted models (LightGBM and XGBoost) and