Bio

Jassim received his bachelor of technology degree in computer science and engineering from the University of Calicut, India. He has 10 years of work experience. He started his career working for the software consulting firm Wipro Technologies in different technology areas like core mobile network development, millicode development for a NonStop OS Kernel, big data and high-performance computing design and development. He later moved on to work with a couple of successful startups on cancer genomics and HR analytics, to design and develop data algorithms and ML data pipelines, respectively. Currently, he is working with Vodafone as their senior data scientist within the telecom business intelligence and finance domain. His technical interests are data science, machine learning, artificial intelligence, natural language processing, block-chain technologies and data visualization. He is currently working on AI-based conversational UI solutions for telecom business intelligence.

In his free time, Jassim enjoys reading, traveling and photography.

Project: Churn Classification for Mobile Telecom CDR Data Using a Neural Network in the Wolfram Language

Goal of the project:

Churn in the telecommunication industry happens when customers leave the current brand and move to another telecom company. With the increasing number of churns, it becomes the operator’s process to retain the profitable customers, known as churn management. In the communication service provider (CSP) industry, each company provides the customers with huge incentives to lure them to switch to their company. The approach here is to aggregate the data for the required analysis and classify potential customers who might churn. The outcome can be used for various business use cases to improve targeted marketing, improve product design, identify the network fault that led to churn and detect potential fraud. With the impending risk of OTT VoIP cannibalization, the clear understanding of their customers’ behavior toward churn is vital for the business to maintain their steady revenue stream.

Summary of work:

The approach here is to aggregate the CDR data for the required analysis with customer call details with a list of churners and non-churners. Divide the dataset into training and testing datasets in an 80:20 ratio. Build a simple neural network and train it using the training dataset to classify potential customers who might churn. Test the model on the test dataset.

Results and future work:

Test results outcomes were fair. Future work would include: (1) validating the model on real unclassified data using GPU for the neural network; and (2) improving understanding and building a model using the deep learning method to predict customer churn in a mobile telecommunication network.