Improving Customer Churn Prediction for OTT Platforms with ‎Machine Learning

Delivering music, video, and media material online without depending on conventional cable or satellite ‎providers is known as over-the-top, or OTT. These services remove the need for pricey contracts and ‎provide a more affordable option for content access. Through a website or app, users can register to ‎watch movies, TV series, and on-demand content. Features like saved favorites, tailored suggestions, ‎and access to unique material are frequently found on OTT platforms. Given how frequently consumers ‎cancel their subscriptions, subscriber churn is a major problem for OTT companies. Because it affects ‎future revenue and service duration, it has a major impact on customer lifetime value. A churn prediction system is required to forecast client attrition to address this issue. Predictive modeling is a ‎useful technique for churn prediction because of machine learning, which enables businesses to proactively handle customer attrition. This paper is the primary contribution to the partial churn definition. This ‎uses machine learning to find commonalities between churners and active customers. Furthermore, the ‎hybrid feature selection method identifies the most significant predictive elements in an actual dataset. ‎Tests on the Kaggle churn modeling dataset show that the suggested framework outperforms other ma-‎machine learning models with an accuracy rate of 98%‎.