MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202611111906 A) filed by Krishna Institute Of Engineering & Technology Kiet on September 17, 2026, for System And Method For Explainable Churn Prediction And Attribution-Driven Generative Retention Response.
Inventors include Aditi Kaushik; and Jyoti.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: Disclosed is a processor-executed churn prediction and retention message generation system and method. A data-intake module receives and structures a plurality of subscriber records into a feature matrix. A balancing module applies a synthetic minority over-sampling operation to the feature matrix to generate a balanced training set. A prediction module, comprising a trained ensemble classifier trained on the balanced training set, generates a churn indication from a feature vector for a subscriber. An attribution module computes a plurality of feature contribution values for the subscriber and ranks a plurality of features according thereto to generate a ranked set of drivers. A generative module receives the ranked set of drivers as a structured conditioning input and generates therefrom an individualised retention message addressing the ranked set of drivers, thereby coupling the ranked set of drivers to the generated retention message.
Disclaimer: Curated by HT Syndication.
