Evaluative study of cluster based customer churn prediction against conventional RFM based churn model

Churn prediction in retail industry is a very critical measure as retention strategies play a crucial role in any successful business model. In the era of e-commerce and customized customer solutions, it is essential for businesses to be on their toes to predict the changes in customer behavior as early as possible in their life cycle. RFM analysis is one of the age old techniques that have proven to be an accurate measure of customer behavior and although it aids in customized targeting, it also requires time to amass enough data. It sometimes means that by the time the business collects substantial data about a customer to make strategic decisions, they are already half way through their change in interests or behavior. Therefore, it is essential to study the possibilities of predicting future states of a customer well in advance. This study focuses on evaluating the accuracy of Machine Learning Models in predicting the churn of retail customers based on their k-means clusters against their actual RFM features. This evaluative comparison would help to establish how reliable it is to utilize a model's results to feed another predictive model and its further applications.