Practical Implementation of Machine Learning and Predictive Analytics in Cellular Network Transactions in Real Time

EasyChair Preprint 683

Abstract

In order to keep a high revenue stream, Communication Service Providers in general, Network Mobile Operators specifically need to ensure a good level of customer satisfaction by assigning a big weight on the user’s Quality of Experience (QoE). With billions of transactions done by customers on both voice and data daily, Communication Service Providers (CSPs) shift the focus in studying customer behavior and data patterns to pinpoint opportunities to improve customer services, service quality and predict when customers are likely to terminate contracts, to perhaps move to another CSP. CSPs have managed to build efficient IT infrastructures to store customer transactions. These exist in many forms such as file systems, databases, etc. In this paper, a simplified predictive analytics is done using the (Customer Relationship Management) CRM information records to classify potential customers likely to terminate their contracts, using logistic regression and random forest models. The paper describes the process to build a simple predictive models to apply on a telecoms dataset.

Keyphrases: Artificial Intelligence AI, CRM, CSP, Predictive Analytics, Random Forest, Telecommunications, logistic regression, machine learning

@booklet{EasyChair:683,
  author    = {Dahj Muwawa Jean Nestor and Kingsley A. Ogudo},
  title     = {Practical  Implementation of Machine Learning and Predictive Analytics in Cellular Network Transactions in Real Time},
  howpublished = {EasyChair Preprint 683},
  year      = {EasyChair, 2018}}