Customer churn is an essential retail metric used in business predictive analytics systems to quantify the number of customers who left a company. All retail and business to consumer companies carefully analyze customer behavior to prevent them to cease their relationship with the company, in other words to make churn. With the latest advancements of Artificial Intelligence and particularly related to Deep Learning, we have a new set of powerful tools ready to employ within multiple horizontal and vertical domain – such as the horizontal of predictive business analytics domain. One of the main goals of predictive analytics is the research and development of the almost-perfect churn detection system. This paper objective is to propose a state-of-the-art churn prediction model based on deep neural models, time-to-next-event models and employing Big Data processing on massive parallel computing using GPU cells.
Keywords — machine learning, business predictive analytics, massive parallel computing, on GPU, deep learning, customer retention, big data, churn prediction
Reference – https://ieeexplore.ieee.org/document/8514153