Customer Behaviour Analysis Using Big Data Analytics

The expansion of digital commerce and online engagement has resulted in large amounts of customer information; thus allowing and challenging present businesses with managing this data. This paper includes a comprehensive structure for analyzing customer behaviour with Big Data Analytics and Machine Learning (ML). We consider five main problems within marketing: customer lifetime value (CLV) prediction, identification of prospects with a high probability of purchase, selecting the best communication channel, predicting customer churn, and performing sentiment analysis. A practical model to predict CLV has been created and tested against a real world e-commerce dataset consisting of 397,925 transactions for 2,845 unique customers using BG-NBD and Gamma-Gamma probabilistic models. The successfully tested model achieved a prediction accuracy of 91.4 percent and indicated a 23:1 ratio of CLV between the top tier and bottom tier segments of customers. The results from comparative analysis demonstrated that both ensemble methods and probabilistic models are superior to traditional rule-based methods for all five use cases. Overall, these findings provide practical information for marketers and data scientists who wish to utilize big data technology strategically for competitive customer relationship management.