Integrated Customer Analytics using Explainability and AutoML for Telecommunications

Abstract

This research study provides an outlook on modeling an integrated customer analytic framework using complex black-box AutoML pipelines while providing insights and explanations to the predictions provided to the customer data mainly in two use cases: Customer Churn and Customer Segmentation. Upon the Literature Review conducted, a pipeline has been derived to integrate both the use cases using supervised and unsupervised models, and explanations were obtained using XAI techniques. The experiments were conducted on the model with desired results and fairness checks were done to check the integrity of the predictions and explanations. The purpose of this research study is to automate the customer analysis process with a comparatively better performance without building a manual pipeline from scratch.