Case Study: Cognizant speeds telecom customer churn analysis and cuts development time with MathWorks (MATLAB)

Cognizant Speeds Customer Churn Analysis for Telecom Service Provider

Cognizant was hired by a major telecom provider to analyze more than 500 GB of customer data and develop predictive models to forecast churn and identify its principal drivers. To handle the large, mixed-format data and meet tight deadlines, Cognizant used MathWorks' MATLAB platform and its toolboxes (Database, Parallel Computing, Deep Learning, and Statistics and Machine Learning).

Using MathWorks' MATLAB and toolboxes, Cognizant cleaned and preprocessed the data, stored and joined tables via Database Toolbox, accelerated processing with Parallel Computing Toolbox, and evaluated neural networks, decision trees, and logistic regression—finding logistic regression most accurate. The MathWorks-based solution identified delayed responses, delayed delivery, and quality issues as top churn drivers, cut algorithm development time from weeks to days, completed the project in less than half the allotted time, and enabled analysis of hundreds of gigabytes of data.

Cognizant

G Subrahmanya VRK Rao

Senior Director of Technology